% This manual is copyright (C) 1984 by the American Mathematical Society. % All rights are reserved! % The file is distributed only for people to see its examples of TeX input, % not for use in the preparation of books like The TeXbook. % Permission for any other use of this file must be obtained in writing % from the copyright holder and also from the publisher (Addison-Wesley). \loop\iftrue \errmessage{This manual is copyrighted and should not be TeXed}\repeat \pausing1 \input manmac \ifproofmode\message{Proof mode is on!}\pausing1\fi % halftitle \titlepage \pageno=-1983 \line{\cmman The\hfill T\kern-10pt\lower13pt\hbox{E}\kern-5pt Xbook} \vfill \ifproofmode \rightline{The fine print in the upper right-hand} \rightline{corner of each page is a draft of intended} \rightline{index entries; it won't appear in the real book.} \rightline{Some index entries will be in |typewriter type|} \rightline{and/or preceded by {\tt\char`\\} or enclosed in \<$\ldots$>, etc;} \rightline{such typographic distinctions aren't shown here.} \rightline{An index entry often extends for several pages;} \rightline{the actual scope will be determined later.} \rightline{Please note things that should be indexed but aren't.} \fi \eject % title \pageno=-1 % the front matter is numbered with roman numerals \font\auth=cmssdc10 scaled\magstep4 % used only on the title page \font\elevenbf=cmbx10 scaled\magstephalf % ditto \font\elevenit=cmti10 scaled\magstephalf % ditto \font\elevenrm=cmr10 scaled\magstephalf % ditto \titlepage \line{\cmman The\hfill T\kern-10pt\lower13pt\hbox{E}\kern-5pt Xbook} ^^{Knuth, Donald Ervin} ^^{Bibby, Duane Robert} \vskip 1pc \baselineskip 13pt \elevenbf \halign to\hsize{#\hfil\tabskip 0pt plus 1fil&#\hfil\tabskip0pt\cr \kern5.5mm\auth DONALD \kern-1pt E. \kern-1pt KNUTH& \elevenit Stanford University\cr \noalign{\vskip 12pc} &\elevenit I\kern.7ptllustrations by\cr &DU\kern-1ptANE BIBBY\cr \noalign{\vfill} &\setbox0=\hbox{\manual77}% \setbox2=\hbox to\wd0{\hss\manual6\hss}% \raise2.3mm\box2\kern-\wd0\box0\cr % A-W logo &ADDISON\kern.1em--WESLEY\cr %&PUBLISHING COMP\kern-.13emANY\kern-1.5mm\cr \noalign{\vskip.5pc \global\elevenrm} &Upper Saddle River, NJ\cr &Boston\enspace$\cdot$\enspace Indianapolis\cr &San Francisco\enspace$\cdot$\enspace New York\cr &Toronto\enspace$\cdot$\enspace Montr\'eal\cr &London\enspace$\cdot$\enspace Munich\cr &Paris\enspace$\cdot$\enspace Madrid\cr &Capetown\enspace$\cdot$\enspace Sydney\enspace$\cdot$\enspace Tokyo\cr &Singapore\enspace$\cdot$\enspace Mexico City\cr} \eject % copyright \titlepage \eightpoint \vbox to 8pc{} \noindent\strut This manual describes \TeX\ Version 3.0. Some of the advanced features mentioned here are absent from earlier versions. \medskip \noindent The quotation on page \sesame\ is copyright $\copyright$ 1970 by Sesame Street, Inc., and used by permission of the Children's Television Workshop. \medskip \noindent \TeX\ is a trademark of the American Mathematical Society. \medskip \noindent {\manual opqrstuq} is a trademark of Addison\kern.1em--Wesley Publishing Company. \bigskip\medskip \noindent {\bf Library of Congress cataloging in publication data} \medskip {\tt\halign{#\hfil\cr Knuth, Donald Ervin, 1938-\cr \ \ \ The TeXbook.\cr \noalign{\medskip} \ \ \ (Computers \& Typesetting ; A)\cr \ \ \ Includes index.\cr \ \ \ 1.~TeX (Computer system).\ \ 2.~Computerized\cr typesetting.\ \ 3.~Mathematics printing.\ \ I.~Title.\cr II.\ Series:\ \ Knuth, Donald Ervin, 1938-\ \ \ \ .\cr Computers \& typesetting ; A.\cr Z253.4.T47K58\ \ 1986\ \ \ \ \ \ \ \ \ 686.2\char13 2544\ \ \ \ \ \ 85-30845\cr ISBN 0-201-13447-0\cr ISBN 0-201-13448-9 (soft)\cr}} \vfill \noindent %{\sl \kern-1pt First hardcover edition, January 1986} {\sl \kern-1pt Incorporates the final corrections made in 1996, and a few dozen more.} \smallskip \noindent Internet page {\tt http://www-cs-faculty.stanford.edu/\char`\~ knuth/abcde.html} contains current information about this book and related books. \smallskip \noindent Copyright $\copyright$ 1984, 1986 by the American Mathematical Society \smallskip \noindent This book is published jointly by the American Mathematical Society and Addison\kern.1em--Wesley Publishing Company. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior written permission of the publishers. Printed in the United States of America. % Published simultaneously in Canada. \medskip \noindent %ISBN 0-201-13448-9\par % paperback ISBN 0-201-13447-0\par % hardcover \noindent %33 34 35 36 37 38 39 DOC 09 08 07 06 % paperback 18 19 20 21 22 23 24 DOC 10 09 08 07 06 % hardcover ^^{Knuth, Donald Ervin} ^^|\copyright| \eject % dedication \titlepage \vbox to 8pc{} \rightline{\strut\eightssi To Jill:} ^^{Knuth, Jill Carter} \vskip2pt \rightline{\eightssi For your books and brochures} \vfill \eject % blank page \titlepage \null\vfill \eject % the preface \titlepage \def\rhead{Preface} \vbox to 8pc{ \rightline{\titlefont Preface}\vss} {\topskip 9pc % this makes equal sinkage throughout the Preface \vskip-\parskip \tenpoint \noindent\hang\hangafter-2 \smash{\lower12pt\hbox to 0pt{\hskip-\hangindent\cmman G\hfill}}\hskip-16pt {\sc ENTLE} R{\sc EADER}: \strut This is a handbook about \TeX, a new typesetting system intended for the creation of beautiful books---and especially for books that contain a lot of mathematics. By preparing a manuscript in \TeX\ format, you will be telling a computer exactly how the manuscript is to be transformed into pages whose typographic quality is comparable to that of the world's finest printers; yet you won't need to do much more work than would be involved if you were simply typing the manuscript on an ordinary typewriter. In fact, your total work will probably be significantly less, if you consider the time it ordinarily takes to revise a typewritten manuscript, since computer text files are so easy to change and to reprocess. \ (If such claims sound too good to be true, keep in mind that they were made by \TeX's designer, on a day when \TeX\ happened to be working, so the statements may be biased; but read on anyway.) This manual is intended for people who have never used \TeX\ before, as well as for experienced \TeX\ hackers. In other words, it's supposed to be a panacea that satisfies everybody, at the risk of satisfying nobody. Everything you need to know about \TeX\ is explained here somewhere, and so are a lot of things that most users don't care about. If you are preparing a simple manuscript, you won't need to learn much about \TeX\ at all; on the other hand, some things that go into the printing of technical books are inherently difficult, and if you wish to achieve more complex effects you will want to penetrate some of \TeX's darker corners. In order to make it possible for many types of users to read this manual effectively, a special sign is used to designate material that is for wizards only: When the symbol $$\vbox{\hbox{\dbend}\vskip 11pt}$$ appears at the beginning of a paragraph, it warns of a ``^{dangerous bend}'' in the train of thought; don't read the paragraph unless you need to. Brave and experienced drivers at the controls of \TeX\ will gradually enter more and more of these hazardous areas, but for most applications the details won't matter. All that you really ought to know, before reading on, is how to get a file of text into your computer using a standard editing program. This manual explains what that file ought to look like so that \TeX\ will understand it, but basic computer usage is not explained here. Some previous experience with technical typing will be quite helpful if you plan to do heavily mathematical work with \TeX, although it is not absolutely necessary. \TeX\ will do most of the necessary formatting of equations automatically; but users with more experience will be able to obtain better results, since there are so many ways to deal with formulas. Some of the paragraphs in this manual are so esoteric that they are rated $$\vcenter{\hbox{\dbend\kern1pt\dbend}\vskip 11pt}\;;$$ everything that was said about single dangerous-bend signs goes double for these. You should probably have at least a month's experience with \TeX\ before you attempt to fathom such doubly dangerous depths of the system; in fact, most people will never need to know \TeX\ in this much detail, even if they use it every day. After all, it's possible to drive a car without knowing how the engine works. Yet the whole story is here in case you're curious. \ (About \TeX, not cars.) The reason for such different levels of complexity is that people change as they grow accustomed to any powerful tool. When you first try to use \TeX, you'll find that some parts of it are very easy, while other things will take some getting used to. A day or so later, after you have successfully typeset a few pages, you'll be a different person; the concepts that used to bother you will now seem natural, and you'll be able to picture the final result in your mind before it comes out of the machine. But you'll probably run into challenges of a different kind. After another week your perspective will change again, and you'll grow in yet another way; and so on. As years go by, you might become involved with many different kinds of typesetting; and you'll find that your usage of \TeX\ will keep changing as your experience builds. That's the way it is with any powerful tool: There's always more to learn, and there are always better ways to do what you've done before. At every stage in the development you'll want a slightly different sort of manual. You may even want to write one yourself. By paying attention to the dangerous bend signs in this book you'll be better able to focus on the level that interests you at a particular time. Computer system manuals usually make dull reading, but take heart: This one contains {\sc ^{JOKES}} every once in a while, so you might actually enjoy reading it. \ (However, most of the jokes can only be appreciated properly if you understand a technical point that is being made---so read {\sl carefully}.) Another noteworthy characteristic of this manual is that it doesn't always tell the ^{truth}. When certain concepts of \TeX\ are introduced informally, general rules will be stated; afterwards you will find that the rules aren't strictly true. In general, the later chapters contain more reliable information than the earlier ones do. The author feels that this technique of deliberate lying will actually make it easier for you to learn the ideas. Once you understand a simple but false rule, it will not be hard to supplement that rule with its exceptions. In order to help you internalize what you're reading, {\sc ^{EXERCISES}} are sprinkled through this manual. It is generally intended that every reader should try every exercise, except for questions that appear in the ``dangerous bend'' areas. If you can't solve a problem, you can always look up the answer. But please, try first to solve it by yourself; then you'll learn more and you'll learn faster. Furthermore, if you think you do know the solution, you should turn to Appendix~A and check it out, just to make sure. The \TeX\ language described in this book is similar to the author's first attempt at a document formatting language, but the new system differs from the old~one in literally thousands of details. Both languages have been called \TeX; but henceforth the old language should be called \TeX78, and its use should rapidly fade away. Let's keep the name \TeX\ for the language described here, since it is so much better, and since it is not going to change any more. ^^{TeX78} I wish to thank the hundreds of people who have helped me to formulate this ``definitive edition'' of the \TeX\ language, based on their experiences with preliminary versions of the system. My work at Stanford has been generously supported by the ^{National Science Foundation}, the ^{Office of Naval Research}, the ^{IBM Corporation}, and the ^{System Development Foundation}. I also wish to thank the ^{American Mathematical Society} for its encouragement, for establishing the \TeX\ Users Group, and for publishing the {\sl ^{TUGboat}\/} newsletter (see Appendix~J). \medskip \line{{\sl Stanford, California}\hfil--- D. E. K.}^^{Knuth, Don} \line{\sl June 1983\hfil} } % end of the special \topskip \endchapter `Tis pleasant, sure, to see one's name in print; A book's a book, although there's nothing in 't. \author ^{BYRON}, {\sl English Bards and Scotch Reviewers\/} (1809) \bigskip A question arose as to whether we were covering the field that it was intended we should fill with this manual. \author RICHARD R. ^{DONNELLEY}, {\sl Proceedings, United % Typothet{\ae} of America\/} (1897) \eject % the table of contents \titlepage \vbox to 8pc{ \rightline{\titlefont Contents} \vfill} ^^{Contents of this manual, table} \def\rhead{Contents} \tenpoint \begingroup \countdef\counter=255 \def\diamondleaders{\global\advance\counter by 1 \ifodd\counter \kern-10pt \fi \leaders\hbox to 20pt{\ifodd\counter \kern13pt \else\kern3pt \fi .\hss}} \baselineskip 15pt plus 5pt \def\\#1. #2. #3.{\line{\strut \hbox to\parindent{\bf\hbox to 1em{\hss#1}\hss}% \rm#2\diamondleaders\hfil\hbox to 2em{\hss#3}}} \\1. The Name of the Game. 1. \\2. Book Printing versus Ordinary Typing. 3. \\3. Controlling \TeX. 7. \\4. Fonts of Type. 13. \\5. Grouping. 19. \\6. Running \TeX. 23. \\7. How \TeX\ Reads What You Type. 37. \\8. The Characters You Type. 43. \\9. \TeX's Roman Fonts. 51. \\10. Dimensions. 57. \\11. Boxes. 63. \\12. Glue. 69. \\13. Modes. 85. \\14. How \TeX\ Breaks Paragraphs into Lines. 91. \\15. How \TeX\ Makes Lines into Pages. 109. \\16. Typing Math Formulas. 127. \\17. More about Math. 139. \\18. Fine Points of Mathematics Typing. 161. \\19. Displayed Equations. 185. \\20. Definitions (also called Macros). 199. \\21. Making Boxes. 221. \\22. Alignment. 231. \\23. Output Routines. 251. \eject \vbox to 8pc{} \\24. Summary of Vertical Mode. 267. \\25. Summary of Horizontal Mode. 285. \\26. Summary of Math Mode. 289. \\27. Recovery from Errors. 295. \null \leftline{\indent\bf Appendices} \\A. Answers to All the Exercises. 305. \\B. Basic Control Sequences. 339. \\C. Character Codes. 367. \\D. Dirty Tricks. 373. \\E. Example Formats. 403. \\F. Font Tables. 427. \\G. Generating Boxes from Formulas. 441. \\H. Hyphenation. 449. \\I\hskip 1pt. Index. 457. \\J\hskip 1pt. Joining the \TeX\ Community. 483. \null % 17 lines so far to balance the 23 on the other page \null % 18 \null % 19 \null % 20 \null % 21 \null % 22 \null % 23 \eject \endgroup \beginchapter Chapter 1. The Name of\\the Game \pageno=1 % This is page number 1, number 1, English words like `technology' stem from a Greek root beginning with the letters $\tau\epsilon\chi\ldots\,$; and this same Greek word means {\sl art\/} as well as technology. Hence the name \TeX, which is an uppercase form of $\tau\epsilon\chi$.^^{TeX (actually \TeX), meaning of} ^^|\tau|^^|\epsilon|^^|\chi| Insiders pronounce the $\chi$ of \TeX\ as a Greek chi, not as an `x', so that \TeX\ rhymes with the word blecchhh. It's the `ch' sound in Scottish words like {\sl loch\/} or German words like {\sl ach\/}; it's a Spanish `j' and a Russian `kh'. When you say it correctly to your computer, the terminal may become slightly moist. The purpose of this pronunciation exercise is to remind you that \TeX\ is primarily concerned with high-quality technical manuscripts: Its emphasis is on art and technology, as in the underlying Greek word. If you merely want to produce a passably good document---something acceptable and basically readable but not really beautiful---a simpler system will usually suffice. With \TeX\ the goal is to produce the {\sl finest\/} quality; this requires more attention to detail, but you will not find it much harder to go the extra distance, and you'll be able to take special pride in the finished product. ^^{beauty} % since truth is indexed, ... On the other hand, it's important to notice another thing about \TeX's name: The `E' is out of kilter. This ^^{logo} displaced `E' is a reminder that \TeX\ is about typesetting, and it distinguishes \TeX\ from other system names. In fact, ^{TEX} (pronounced {\sl tecks\/}) is the admirable {\sl Text EXecutive\/} processor developed by ^{Honeywell Information Systems}. Since these two system names are ^^{Bemer, Robert, see TEX, ASCII} pronounced quite differently, they should also be spelled differently. The correct way to refer to \TeX\ in a computer file, or when using some other medium that doesn't allow lowering of the `E', is to type `^|TeX|'. Then there will be no confusion with similar names, and people will be primed to pronounce everything properly. \exercise After you have mastered the material in this book, what will you be: a \TeX pert, or a \TeX nician? \answer A \TeX nician (underpaid); sometimes also called a \TeX acker. \endchapter They do certainly give very strange and new-fangled names to diseases. \author ^{PLATO}, {\sl The Republic}, Book 3 (c.\ 375 B.C.) % 405c \bigskip Technique! The very word is like the shriek Of outraged Art. It is the idiot name Given to effort by those who are too weak, Too weary, or too dull to play the game. \author LEONARD ^{BACON}, {\sl Sophia Trenton\/} (1920) % composed at Stanford \eject \beginchapter Chapter 2. Book Printing\\versus\\Ordinary Typing When you first started using a computer terminal, you probably had to adjust to the difference between the digit `1' and the lowercase letter `l'. When you take the next step to the level of typography that is common in book publishing, a few more adjustments of the same kind need to be made; your eyes and your fingers need to learn to make a few more distinctions. In the first place, there are two kinds of ^{quotation marks} in books, but only one kind on the typewriter. Even your computer terminal, which has more characters than an ordinary typewriter, probably has only a non-oriented double-quote mark (|"|), because the standard ^{ASCII} code for computers was not invented with book publishing in mind. However, your terminal probably does have two flavors of single-quote marks, namely |`| and |'|; the second of these is useful also as an ^{apostrophe}. American keyboards usually contain a left-quote character that shows up as something like {\tt\char'22}, and an apostrophe or right-quote that looks like {\tt\char'15} or {\tt\char'23}. To produce double-quote marks with \TeX, you simply type two single-quote marks of the appropriate kind. For example, to get the phrase \begindisplay ``I understand.'' \enddisplay (including the quotation marks) you should type \begintt ``I understand.'' \endtt to your computer. A typewriter-like style of type will be used throughout this manual to indicate \TeX\ constructions that you might type on your terminal, so that the symbols actually typed are readily distinguishable from the output \TeX\ would produce and from the comments in the manual itself. Here are the symbols to be used in the examples: \begintt ABCDEFGHIJKLMNOPQRSTUVWXYZ abcdefghijklmnopqrstuvwxyz 0123456789"#$%&@*+-=,.:;?! ()<>[]{}`'\||/_^~ \endtt If your computer terminal doesn't happen to have all of these, don't despair; \TeX\ can make~do with the ones you have. An additional symbol \begindisplay \] \enddisplay is used to stand for a {\sl ^{blank space}}, in case it is important to emphasize that a blank space is being typed; thus, what you {\sl really\/} type in the example above is \begintt ``I|]understand.'' \endtt Without such a symbol you would have difficulty seeing the invisible parts of certain constructions. But we won't be using `\]' very often, because spaces are usually visible enough. Book printing differs significantly from ordinary typing with respect to ^{dashes}, ^{hyphens}, and ^{minus signs}. In good math books, these symbols are all different; in fact there usually are at least four different symbols: \begindisplay a hyphen (-);\cr an en-dash (--);\cr an em-dash (---);\cr a minus sign ($-$).\cr \enddisplay Hyphens are used for compound words like `daughter-in-law' and `X-rated'. ^{En-dash}es are used for number ranges like `pages 13--34', and also in contexts like `exercise 1.2.6--52'. ^{Em-dash}es are used for punctuation in sentences---they are what we often call simply dashes. And minus signs are used in formulas. A conscientious user of \TeX\ will be careful to distinguish these four usages, and here is how to do it: \begindisplay for a hyphen, type a hyphen (|-|);\cr for an en-dash, type two hyphens (|--|);\cr for an em-dash, type three hyphens (|---|);\cr for a minus sign, type a hyphen in mathematics mode (|$-$|).\cr \enddisplay (Mathematics mode occurs between dollar signs; it is discussed later, so you needn't worry about it now.) \exercise Explain how to type the following sentence to \TeX: Alice said, ``I always use an en-dash instead of a hyphen when specifying page numbers like `480--491' in a ^{bibliography}.'' \answer |Alice said, ``I always use an en-dash instead of a hyphen when|\break |specifying page numbers like `480--491' in a bibliography.''| \ (The wrong answer to this question ends with |'480-49l' in a bibliography."|) \exercise What do you think happens when you type four hyphens in a row? \answer You get em-dash and hyphen (----), which looks awful. If you look closely at most well-printed books, you will find that certain combinations of letters are treated as a unit. For example, this is true of the `f' and the `i' of `find'. Such combinations are called {\sl ^{ligatures}}, and professional typesetters have traditionally been trained to watch for letter combinations such as |ff|, |fi|, |fl|, |ffi|, and |ffl|. \ (The reason is that words like `f{}ind' don't look very good in most styles of type unless a ligature is substituted for the letters that clash. It's somewhat surprising how often the traditional ligatures appear in English; other combinations are important in other languages.) \exercise Think of an English word that contains two ligatures. \answer fluffier firefly fisticuffs, flagstaff fireproofing, chiffchaff and riffraff. The good news is that you do {\sl not\/} have to concern yourself with ligatures: \TeX\ is perfectly capable of handling such things by itself, using the same mechanism that converts `|--|' into `--'. In fact, \TeX\ will also look for combinations of adjacent letters (like `|A|' next to `|V|'\thinspace) that ought to be moved closer together for better appearance; this is called {\sl ^{kerning}}. \medbreak To summarize this chapter: When using \TeX\ for straight copy, you type the copy as on an ordinary typewriter, except that you need to be careful about quotation marks, the number 1, and various kinds of hyphens/dashes. \TeX\ will automatically take care of other niceties like ligatures and kerning. \danger (Are you sure you should be reading this paragraph? The ``^{dangerous bend}'' sign here is meant to warn you about material that ought to be skipped on first reading. And maybe also on second reading. The reader-beware paragraphs sometimes refer to concepts that aren't explained until later chapters.) \danger If your keyboard does not contain a left-quote symbol, you can type ^|\lq|, followed by a space if the next character is a letter, or followed by a |\| if the next character is a space. Similarly, ^|\rq| yields a right-quote character. Is that clear? \begintt \lq\lq|]I|]understand.\rq\rq\|] \endtt \danger In case you need to type ^{quotes within quotes}, for example a single quote followed by a double quote, you can't simply type \thinspace|'''|\thinspace\ because \TeX\ will interpret this as ''' (namely, double quote followed by single quote). If you have already read Chapter~5, you might expect that the solution will be to use grouping---namely, to type something like \thinspace|{'}''|. But it turns out that this doesn't produce the desired result, because there is usually less space following a single right quote than there is following a double right quote: What you get is {'}'', which is indeed a single quote followed by a double quote (if you look at it closely enough), but it looks almost like three equally spaced single quotes. On the other hand, you certainly won't want to type \thinspace|'|\]|''|, because that space is much too large---it's just as large as the space between words---and \TeX\ might even start a new line at such a space when making up a paragraph! The solution is to type \thinspace|'\thinspace''|, which produces '\thinspace'' as desired.^^|\thinspace| \dangerexercise OK, now you know how to produce ''' and '\thinspace''; how do you get ``\thinspace` and `{}``\thinspace? \answer |``\thinspace`|; and either |`{}``| or |{`}``| or something similar. Reason: There's usually less space {\sl preceding\/} a single left quote than there is preceding a double left quote. \ (Left and right are opposites.) \dangerexercise Why do you think the author introduced the control sequence |\thinspace| to solve the adjacent-quotes problem, instead of recommending the trickier construction |'$|^|\,||$''| (which also works)? \answer Eliminating ^|\thinspace| would mean that a user need not learn the term; but it is not advisable to minimize terminology by ``overloading'' math mode with tricky constructions. For example, a user who wishes to take advantage of \TeX's ^|\mathsurround| feature would be thwarted by non-mathematical uses of dollar signs. \ (Incidentally, neither |\thinspace| nor ^|\,| are built into \TeX; both are defined in terms of more primitive features, in Appendix~B.) \endchapter In modern Wit all printed Trash, is Set off with num'rous\/ {\rm Breaks}\raise.5ex\vbox{\hrule width 2em}% and\/ {\rm Dashes}\raise.5ex\vbox{\hrule width 1em} % no period after the em-dash: stet! % Sir Walter Scott ruined this quote in his edition of Swift! \author JONATHAN ^{SWIFT}, {\sl On Poetry: A Rapsody\/} (1733) % lines 93--94 % Rapsody: stet! \bigskip Some compositors still object to work in offices where type-composing machines are introduced. \author WILLIAM STANLEY ^{JEVONS}, {\sl Political Economy\/} (1878) % sec 55 % "They are all afraid that if the work is done too easily and rapidly, % they will not be wanted to do it." % Jevons goes on to say that justifying and page makeup can't be done % profitably by machines, so the employees needn't fear losing their jobs. \eject \beginchapter Chapter 3. Controlling\\\TeX Your keyboard has very few keys compared to the large number of symbols that you may want to specify. In order to make a limited keyboard sufficiently versatile, one of the characters that you can type is reserved for special use, and it is called the {\sl ^{escape character}}. Whenever you want to type something that controls the format of your manuscript, or something that doesn't use the keyboard in the ordinary way, you should type the escape character followed by an indication of what you want to do. Note: Some computer terminals have a key marked `|ESC|', but that is {\sl not\/} your escape character! It is a key that sends a special message to the operating system, so don't confuse it with what this manual calls ``escape.'' \TeX\ allows any character to be used for escapes, but the ``^{backslash}'' character `|\|' is usually adopted for this purpose, since backslashes are reasonably convenient to type and they are rarely needed in ordinary text. Things work out best when different \TeX\ users do things consistently, so we shall escape via backslashes in all the examples of this manual. Immediately after typing `|\|' (i.e., immediately after an escape character) you type a coded command telling \TeX\ what you have in mind. Such commands are called {\sl ^{control sequences}}. For example, you might type ^^{markup commands, see control sequences} \begintt \input MS \endtt which (as we will see later) causes \TeX\ to begin reading a file called `|MS.tex|'; the string of characters `^|\input|' is a control sequence. Here's another example: \begintt George P\'olya and Gabor Szeg\"o. \endtt % sic; this is the spelling used in opening pages of their famous books % but I give the Hungarian spellings in the index \TeX\ converts this to `George P\'olya and Gabor Szeg\"o.' There are two ^^{Polya}^^{Szego}^^{acute}^^{umlaut} control sequences, ^|\'| and ^|\"|, here; these control sequences have been used to place ^{accents} over some of the letters. Control sequences come in two flavors. The first kind, like |\input|, is called a {\sl^{control word}\/}; it consists of an escape character followed by one or more {\sl letters}, followed by a space or by something besides a letter. \ (\TeX\ has to know where the control sequence ends, so you must put a space after a control word if the next character is a letter. For example, if you type `|\inputMS|', \TeX\ will naturally interpret this as a control word with seven letters.) \ In case you're wondering what a ``^{letter}'' is, the answer is that \TeX\ normally regards the 52 symbols |A...Z| and |a...z| as letters. The digits |0...9| are {\sl not\/} considered to be letters, so they don't appear in control sequences of the first kind. A control sequence of the other kind, like |\'|, is called a {\sl ^{control symbol}\/}; it consists of the escape character followed by a single {\sl nonletter}. In this case you don't need a space to separate the control sequence from a letter that follows, since control sequences of the second kind always have exactly one symbol after the escape character. \exercise What are the control sequences in `|\I'm \exercise3.1\\!|'\thinspace? \answer |\I|, |\exercise|, and |\\|. (The last of these is of type~2, i.e., a control symbol, since the second backslash is not a letter; the first backslash keeps the second one from starting its own control sequence.) \exercise We've seen that the input |P\'olya| yields `P\'olya'. Can you guess how the French words `math\'ematique' and `centim\`etre' should be specified? \answer |math\'ematique| and |centim\`etre|.^^|\'|^^|\`| When a space comes after a control word (an all-letter control sequence), it is ignored by \TeX; i.e., it is not considered to be a ``real'' space belonging to the manuscript that is being typeset. But when a space comes after a control symbol, it's truly a space. Now the question arises, what do you do if you actually {\sl want\/} a space to appear after a control word? We will see later that \TeX\ treats two or more consecutive spaces as a single ^{space}, so the answer is {\sl not\/} going to be ``type two spaces.'' The correct answer is to type ``control space,'' ^^|\ | namely \begintt \|] \endtt (the escape character followed by a blank space); \TeX\ will treat this as a space that is not to be ignored. Notice that |\|\] is a control sequence of the second kind, namely a control symbol, since there is a single nonletter (\]) following the escape character. Two consecutive spaces are considered to be equivalent to a single space, so further spaces immediately following |\|\] will be ignored. But if you want to enter, say, three consecutive spaces into a manuscript you can type `|\|\]|\|\]|\|\]'. Incidentally, typists are often taught to put two spaces at the ends of sentences; but we will see later that \TeX\ has its own way to produce extra space in such cases. Thus you needn't be consistent in the number of spaces you type. \danger Nonprinting control characters like \ might follow an escape character, and these lead to distinct control sequences according to the rules. \TeX\ is initially set up to treat |\|\ and |\|\ ^^|\|^^|\| the same as |\|\] (control space); these special control sequences should probably not be redefined, because you can't see the difference between them when you look at them in a file. ^^{carriage-return, see } It is usually unnecessary for you to use ``control space,'' since control sequences aren't often needed at the ends of words. But here's an example that might shed some light on the matter: This manual itself has been typeset by \TeX, and one of the things that occurs fairly often is the tricky ^{logo} `\TeX', which requires backspacing and lowering the E. There's a special control word \begintt \TeX \endtt that produces the half-dozen or so instructions necessary to typeset `\TeX'. When a phrase like `\TeX\ ignores spaces after control words.' is desired, the manuscript renders it as follows: \begintt \TeX\ ignores spaces after control words. \endtt Notice the extra |\| following ^|\TeX|; this produces the control space that is necessary because \TeX\ ignores spaces after control words. Without this extra |\|, the result would have been \begindisplay \TeX ignores spaces after control words. \enddisplay On the other hand, you can't simply put |\| after |\TeX| in all contexts. For example, consider the phrase \begintt the logo `\TeX'. \endtt In this case an extra backslash doesn't work at all; in fact, you get a curious result if you type \begintt the logo `\TeX\'. \endtt Can you guess what happens? \ Answer: The |\'| is a control sequence denoting an acute accent, as in our |P\'olya| example above; the effect is therefore to put an accent over the next nonblank character, which happens to be a period. In other words, you get an accented period, and the result is \begindisplay the logo `\TeX\'. \enddisplay Computers are good at following instructions, but not at reading your mind. \TeX\ understands about 900 control sequences as part of its built-in vocabulary, and all of them are explained in this manual somewhere. But you needn't worry about learning so many different things, because you won't really be needing very many of them unless you are faced with unusually complicated copy. Furthermore, the ones you do need to learn actually fall into relatively few categories, so they can be assimilated without great difficulty. For example, many of the control sequences are simply the names of special characters used in math formulas; you type `^|\pi|'~to get~`$\pi$', `^|\Pi|'~to get~`$\Pi$', `^|\aleph|'~to get~`$\aleph$', `^|\infty|'~to get~`$\infty$', `^|\le|'~to get~`$\le$', `^|\ge|'~to get~`$\ge$', `^|\ne|'~to get~`$\ne$', `^|\oplus|'~to get~`$\oplus$', `^|\otimes|'~to get~`$\otimes$'. Appendix~F contains several tables of such symbols. \danger There's no built-in relationship between ^{uppercase} and ^{lowercase} letters in control sequence names. For example, `|\pi|' and `|\Pi|' and `|\PI|' and `|\pI|' are four different control words. The 900 or so control sequences that were just mentioned actually aren't the whole story, because it's easy to define more. For example, if you want to substitute your own favorite names for math symbols, so that you can remember them better, you're free to go right ahead and do it; Chapter~20 explains how. About 300 of \TeX's control sequences are called {\sl ^{primitive}\/}; these are the low-level atomic operations that are not decomposable into simpler functions. All other control sequences are defined, ultimately, in terms of the primitive ones. For example, ^|\input| is a primitive operation, but ^|\'| and ^|\"| are not; the latter are defined in terms of an ^|\accent| primitive. People hardly ever use \TeX's primitive control sequences in their manuscripts, because the primitives are $\ldots$ well $\ldots$ so {\sl primitive}. You have to type a lot of instructions when you are trying to make \TeX\ do low-level things; this takes time and invites mistakes. It is generally better to make use of higher-level control sequences that state what functions are desired, instead of typing out the way to achieve each function each time. The higher-level control sequences need to be defined only once in terms of primitives. For example, |\TeX| is a control sequence that means ``typeset the \TeX\ logo''; |\'| is a control sequence that means ``put an acute accent over the next character''; and both of these control sequences might require different combinations of primitives when the style of type changes. If \TeX's logo were to change, the author would simply have to change one definition, and the changes would appear automatically wherever they were needed. By contrast, an enormous amount of work would be necessary to change the logo if it were specified as a sequence of primitives each time. At a still higher level, there are control sequences that govern the overall format of a document. For example, in the present book the author typed `^|\exercise|' just before stating each exercise; this |\exercise| command was programmed to make \TeX\ do all of the following things: \nobreak\medskip \item\bull compute the exercise number (e.g., `3.2' for the second exercise in Chapter~3); \smallskip \item\bull typeset `\thinspace{\manual\char'170\rm\kern.15em \ninebf EXERCISE \bf3.2}' with the appropriate typefaces, on a line by itself, and with the triangle sticking out in the left margin; \smallskip \item\bull leave a little extra space just before that line, or begin a new page at that line if appropriate; \smallskip \item\bull prohibit beginning a new page just after that line; \smallskip \item\bull suppress indentation on the following line. \medbreak\noindent It is obviously advantageous to avoid typing all of these individual instructions each time. And since the manual is entirely described in terms of high-level control sequences, it could be printed in a radically different format simply by changing a dozen or so definitions. % and sweating over the page layout in the math and alignment chapters! \danger How can a person distinguish a \TeX\ primitive from a control sequence that has been defined at a higher level? There are two ways: \ (1)~The index to this manual lists all of the control sequences that are discussed, and each primitive is marked with an asterisk. \ (2)~You can display the meaning of a control sequence while running \TeX\null. If you type `^|\show||\cs|' where |\cs| is any control sequence, \TeX\ will respond with its current meaning. For example, `|\show\input|' results in \hbox{`|> \input=\input.|'}, because |\input| is primitive. On the other hand, `|\show|^|\thinspace|' yields \begintt > \thinspace=macro: ->\kern .16667em . \endtt This means that |\thinspace| has been defined as an abbreviation for `|\kern|~|.16667em|~'. By typing `|\show|\penalty0|\kern|' you can verify that ^|\kern| is primitive. The results of\/ |\show| appear on your terminal and in the ^{log file} that you get after running \TeX. \dangerexercise Which of the control sequences |\|\] and |\|\ is primitive? \answer According to the index, |\|\] is primitive but |\|\ isn't. The command `|\def\^^M{\ }|' in Appendix~B is what actually defines |\|\, since a return is representable as |^^M|. Asking \TeX\ to |\show\^^M| \looseness-1 produces the response `|>| |\^^M=macro:->\|\]|.|'. In the following chapters we shall frequently discuss ``^{plain \TeX}'' format, which is a set of about 600 ^{basic control sequences} that are defined in Appendix~B\null. These control sequences, together with the 300 or so primitives, are usually present when \TeX\ begins to process a manuscript; that is why \TeX\ claims to know roughly 900 control sequences when it starts. We shall see how plain \TeX\ can be used to create documents in a flexible format that meets many people's needs, using some typefaces that come with the \TeX\ system. However, you should keep in mind that plain \TeX\ is only one of countless ^{formats} that can be designed on top of \TeX's primitives; if you want some other format, it will usually be possible to adapt \TeX\ so that it will handle whatever you have in mind. The best way to learn is probably to start with plain \TeX\ and to change its definitions, little by little, as you gain more experience. \danger Appendix E contains examples of formats that can be added to Appendix~B for special applications; for example, there is a set of definitions suitable for business correspondence. A complete specification of the format used to typeset this manual also appears in Appendix~E\null. Thus, if your goal is to learn how to design \TeX\ formats, you will probably want to study Appendix~E while mastering Appendix~B\null. After you have become skilled in the lore of control-sequence definition, you will probably have developed some formats that other people will want to use; you should then write a supplement to this manual, explaining your style rules. The main point of these remarks, as far as novice \TeX\ users are concerned, is that it is indeed possible to define nonstandard \TeX\ control sequences. When this manual says that something is part of ``plain \TeX,'' it means that \TeX\ doesn't insist on doing things exactly that way; a person could change the rules by changing one or more of the definitions in Appendix~B\null. But you can safely rely on the control sequences of plain \TeX\ until you become an experienced \TeX nical~typist. \ddangerexercise How many different control sequences of length~2 (including the escape character) are possible? How many of length~3? \answer There are 256 of length~2; most of these are undefined when \TeX\ begins. \ (\TeX\ allows any character to be an escape, but it does not distinguish between control sequences that start with different escape characters.) \ If we assume that there are 52 letters, there are exactly $52^2$ possible control sequences of length~3 (one for each pair of letters, from |AA| to |zz|). But Chapter~7 explains how to use ^|\catcode| to change any character into a ``^{letter}''; therefore it's possible to use any of $256^2$ potential control sequences of length~3. \endchapter Syllables govern the world. \author JOHN ^{SELDEN}, {\sl Table Talk\/} (1689) % section on Power \bigskip I claim not to have controlled events, but confess plainly that events have controlled me. \author ABRAHAM ^{LINCOLN} (1864) % letter to A. G. Hodges, April 4 \eject \beginchapter Chapter 4. Fonts\\of Type Occasionally you will want to change from one ^{typeface} to another, for example if you wish to be {\bf ^{bold}} or to {\sl emphasize\/} something. \TeX\ deals with sets of up to 256 characters called ``^{fonts}'' of type, and control sequences are used to select a particular font. For example, you could specify the last few words of the first sentence above in the following way, using the plain \TeX\ format of Appendix~B: \begintt to be \bf bold \rm or to \sl emphasize \rm something. \endtt Plain \TeX\ provides the following control sequences for changing fonts: \begindisplay ^|\rm| switches to the normal ``roman'' typeface:&Roman\cr ^|\sl| switches to a slanted roman typeface:&\sl Slanted\cr ^|\it| switches to italic style:&\it Italic\cr ^|\tt| switches to a typewriter-like face:&\tt Typewriter\cr ^|\bf| switches to an extended boldface style:&\bf Bold\cr \enddisplay ^^{typewriter type}^^{face} At the beginning of a run you get ^{roman type} (|\rm|) unless you specify otherwise. Notice that two of these faces have an ``^{oblique}'' slope for emphasis: {\sl ^{Slanted type} is essentially the same as roman, but the letters are slightly skewed, \it while the letters in ^{italic type} are drawn in a different style.} \ (You can perhaps best appreciate the difference between the roman and italic styles by contemplating {\tenu letters that are in an unslanted italic face.}) \ Typographic conventions are presently in a state of transition, because new technology has made it possible to do things that used to be prohibitively expensive; people are wrestling with the question of how much to use their new-found typographic freedom. Slanted roman type was introduced in the 1930s, but it first became widely used as an alternative to the conventional italic during the late 1970s. It can be bene\-ficial in mathematical texts, since slanted letters are distinguishable from the italic letters in math formulas. The double use of italic type for two different purposes---for example, when statements of theorems are italicized as well as the names of variables in those theorems---has led to some confusion, which can now be avoided with slanted type. People are not generally agreed about the relative merits of slanted versus italic, but slanted type is rapidly becoming a favorite for the titles of books and journals in bibliographies. Special fonts are effective for emphasis, but not for sustained reading; your eyes would tire if long portions of this manual were entirely set in a bold or slanted or italic face. Therefore roman type accounts for the bulk of most typeset material. But it's a nuisance to say `|\rm|' every time you want to go back to the roman style, so \TeX\ provides an easier way to do it, using ``^{curly brace}^^{brace}'' symbols: You can switch fonts inside the special symbols |{| and |}|, without affecting the fonts outside. For example, the displayed phrase at the beginning of this chapter is usually rendered \begintt to be {\bf bold} or to {\sl emphasize} something. \endtt This is a special case of the general idea of ``^{grouping}'' that we shall discuss in the next chapter. It's best to forget about the first way of changing fonts, and to use grouping instead; then your \TeX\ manuscripts will look more natural, and you'll probably never\footnote*{Well \dots, hardly ever.} have to type `|\rm|'. \exercise Explain how to type the bibliographic reference `Ulrich ^{Dieter}, {\sl Journal f\"ur die reine und angewandte Mathematik\/ \bf201} (1959), 37--70.' [Use grouping.] \answer |Ulrich Dieter, {\sl Journal f\"ur die reine und angewandte|\parbreak |Mathematik\/ \bf201} (1959), 37--70.|\par\nobreak\smallskip\noindent It's convenient to use a single group for both |\sl| and |\bf| here. The `|\/|' is a refinement that you might not understand until you read the rest of Chapter~4. We have glossed over an important aspect of quality in the preceding discussion. Look, for example, at the {\it italicized} and {\sl slanted} words in this sentence. Since italic and slanted styles slope to the right, the d's stick into the spaces that separate these words from the roman type that follows; as a result, the spaces appear to be too skimpy, although they are correct at the base of the letters. To equalize the effective white space, \TeX\ allows you to put the special control sequence `^|\/|' just before switching back to unslanted letters. When you type \begintt {\it italicized\/} and {\sl slanted\/} words \endtt you get {\it italicized\/} and {\sl slanted\/} words that look better. The `|\/|' tells \TeX\ to add an\break % makes the line tighter, to be fair ``{\sl^{italic correction}\/}'' to the previous letter, depending on that letter; this correction is about four times as much for an `$f$' as for a `$c$', in a typical italic font. Sometimes the italic correction is not desirable, because other factors take up the visual slack. The standard rule of thumb is to use |\/| just before switching from slanted or italic to roman or bold, unless the next character is a period or comma. For example, type \begintt {\it italics\/} for {\it emphasis}. \endtt Old manuals of style say that the ^{punctuation} after a word should be in the {\it same\/} font as that {\it word;\/} but an italic semicolon often looks wrong, so this convention is changing. When an italicized word occurs just before a semicolon, the author recommends typing `|{\it word\/};|'. \exercise {\it Explain how to typeset a\/ {\rm roman} word in the midst of an italicized sentence.} \answer |{\it Explain ... typeset a\/ {\rm roman} word ... sentence.}| Note the position of the italic correction in this case. \danger Every letter of every font has an italic correction, which you can bring to life by typing |\/|. The correction is usually zero in unslanted styles, but there are exceptions: To typeset a bold `{\bf f\/}' in quotes, you should say |a| |bold| \hbox{|`{\bf f\/}'|}, lest you get a bold `{\bf f}'. \ddangerexercise Define a control sequence |\ic| such that `|\ic c|' puts the italic correction of character $c$ into \TeX's register |\dimen0|. \answer |\def\ic#1{\setbox0=\hbox{#1\/}\dimen0=\wd0|\parbreak |\setbox0=\hbox{#1}\advance\dimen0 by -\wd0}|. \ddanger The primitive control sequence ^|\nullfont| stands for a font that has no characters. This font is always present, in case you haven't specified any others. Fonts vary in size as well as in shape. For example, the font you are now reading is called a ``10-point'' font, because certain features of its design are 10 ^{points} apart, when measured in printers' units. \ (We will study the point system later; for now, it should suffice to point out that the parentheses around this sentence are exactly 10 points tall---and the em-dash is just 10 points wide.) \ The ``^{dangerous bend}'' sections of this manual are set in 9-point type, the foot\-notes in 8-point, ^{subscripts} in 7-point or 6-point, sub-subscripts in 5-point. Each font used in a \TeX\ manuscript is associated with a control sequence; for example, the 10-point font in this paragraph is called ^|\tenrm|, and the corresponding 9-point font is called ^|\ninerm|. The slanted fonts that match |\tenrm| and |\ninerm| are called ^|\tensl| and ^|\ninesl|. These control sequences are not built into \TeX, nor are they the actual names of the fonts; \TeX\ users are just supposed to make up convenient names, whenever new fonts are introduced into a manuscript. Such control sequences are used to change typefaces. When fonts of different sizes are used simultaneously, \TeX\ will line the letters up according to their ``^{baseline}s.'' For example, if you type \begintt \tenrm smaller \ninerm and smaller \eightrm and smaller \sevenrm and smaller \sixrm and smaller \fiverm and smaller \tenrm \endtt the result is {smaller \ninerm and smaller \eightrm and smaller \sevenrm and smaller \sixrm and smaller \fiverm and smaller}. Of course this is something that authors and readers aren't accustomed to, because printers couldn't do such things with traditional lead types. Perhaps poets who wish to speak in {\fiverm a still small voice} will cause future books to make use of frequent font variations, but nowadays it's only an occasional font freak {\fiverm(like the author of this manual)} who likes such experiments. One should not get too carried away by the prospect of font switching unless there is good reason. An alert reader might well be confused at this point because we started out this chapter by saying that `|\rm|' is the command that switches to roman type, but later on we said that `|\tenrm|' is the way to do it. The truth is that both ways work. But it has become customary to set things up so that |\rm| means ``switch to roman type in the current size'' while |\tenrm| means ``switch to roman type in the 10-point size.'' In plain \TeX\ format, nothing but 10-point fonts are provided, so |\rm| will always get you |\tenrm|; but in more complicated formats the meaning of\/ |\rm| will change in different parts of the manuscript. For example, in the format used by the author to typeset this manual, there's a control sequence `^|\tenpoint|' that causes |\rm| to mean |\tenrm|, |\sl| to mean |\tensl|, and so on, while `^|\ninepoint|' changes the definitions so that |\rm| means |\ninerm|, etc. There's another control sequence used to introduce the quotations at the end of each chapter; when the quotations are typed, |\rm| and |\sl| temporarily stand for {\eightss 8-point unslanted sans-serif type} and {\eightssi 8-point slanted sans-serif type}, respectively. This device of constantly redefining the abbreviations |\rm| and |\sl|, behind the scenes, frees the typist from the need to remember what size or style of type is currently being used. \exercise Why do you think the author chose the names `|\tenpoint|' and `|\tenrm|', etc., instead of `|\10point|' and `|\10rm|'\thinspace? \answer Control word names are made of letters, not digits. \dangerexercise Suppose that you have typed a manuscript using slanted type for emphasis, but your editor suddenly tells you to change all the slanted to italic. What's an easy way to do this? \answer Say |\def\sl{\it}| at the beginning, and delete other definitions of\/ |\sl| that might be present in your format file (e.g., there might be one inside a |\tenpoint| macro). \danger Each font has an external name that identifies it with respect to all other fonts in a particular library. For example, the font in this sentence is called `|cmr9|', which is an abbreviation for ``^{Computer Modern} Roman 9~point.'' ^^{cm fonts} In order to prepare \TeX\ for using this font, the command \begintt \font\ninerm=cmr9 \endtt appears in Appendix E\null. In general you say `^|\font||\cs=|\' to load the information about a particular font into \TeX's memory; afterwards the control sequence |\cs| will select that font for typesetting. Plain \TeX\ makes only sixteen fonts available initially (see Appendix~B and Appendix~F\null), but you can use |\font| to access anything that exists in your system's font library. \danger It is often possible to use a font at several different sizes, by magnifying or shrinking the character images. Each font has a so-called ^{design size}, which reflects the size it normally has by default; for example, the design size of |cmr9| is 9~points. But on many systems there is also a range of sizes at which you can use a particular font, by scaling its dimensions up or down. To load a scaled font into \TeX's memory, you simply say `|\font\cs=|\ ^|at| \'. For example, the command \begintt \font\magnifiedfiverm=cmr5 at 10pt \endtt brings in 5-point Computer Modern Roman at twice its normal size. \ (Caution: Before using this `|at|' feature, you should check to make sure that your typesetter supports the font at the size in question; \TeX\ will accept any \ that is positive and less than 2048 points, but the final output will not be right unless the scaled font really is available on your printing device.) \danger What's the difference between |cmr5| |at| |10pt| and the normal 10-point font, |cmr10|? Plenty; a well-designed font will be drawn differently at different point sizes, and the letters will often have different relative heights and widths, in order to enhance readability. \begindisplay \tenrm Ten-point type is different from% \magnifiedfiverm\ magnif{}ied f{}ive-point type. \enddisplay It is usually best to scale fonts only slightly with respect to their design size, unless the final product is going to be photographically reduced after \TeX\ has finished with it, or unless you are trying for an unusual effect.^^{magnification}^^{reduction} \danger Another way to magnify a font is to specify a scale factor that is relative to the design size. For example, the command \begintt \font\magnifiedfiverm=cmr5 scaled 2000 \endtt is another way to bring in the font ^|cmr5| at double size. The scale factor is specified as an integer that represents a magnification ratio times~1000. Thus, a scale factor of 1200 specifies magnification by 1.2, etc. \dangerexercise State two ways to load font |cmr10| into \TeX's memory at half its normal size. \answer |\font\squinttenrm=cmr10 at 5pt|\parbreak |\font\squinttenrm=cmr10 scaled 500| \font\onerm=cmr10 scaled\magstep1 \font\onett=cmtt10 scaled\magstep1 \font\tworm=cmr10 scaled\magstep2 \font\twott=cmtt10 scaled\magstep2 %\font\threerm=cmr10 scaled\magstep3 % such large magnifications look ugly %\font\threett=cmtt10 scaled\magstep3 % in a book context! \danger At many computer centers it has proved convenient to supply fonts at magnifications that grow in geometric ratios---something like equal-tempered tuning on a ^{piano}. The idea is to have all fonts available at their true size as well as at magnifications 1.2 and~1.44 (which is $1.2\times1.2$); perhaps also at magnification~1.728 ($=1.2\times1.2\times1.2$) and even higher. Then you can magnify an entire document by 1.2 or~1.44 and still stay within the set of available fonts. Plain \TeX\ provides the abbreviations ^|\magstep||0| for a scale factor of 1000, |\magstep1| for a scaled factor of 1200, |\magstep2| for 1440, and so on up to |\magstep5|. You say, for example, \begintt \font\bigtenrm=cmr10 scaled\magstep2 \endtt to load font |cmr10| at $1.2\times1.2$ times its normal size. \begindisplay \lineskip5pt \tenrm\llap{``}This is {\tentt cmr10} at normal size ({\tentt \char`\\magstep0}).''\cr \onerm\llap{``}This is {\onett cmr10} scaled once by 1.2 ({\onett \char`\\magstep1}).''\cr \tworm\llap{``}This is {\twott cmr10} scaled twice by 1.2 ({\twott \char`\\magstep2}).''\cr %\threerm\llap{``}This is {\threett cmr10} scaled by % {\threett \char`\\magstep3}.''\cr \enddisplay (Notice that a little magnification goes a long way.) \ There's also ^|\magstephalf|, which magnifies by $\sqrt{1.2}$, i.e., halfway between steps 0 and~1. \danger Chapter~10 explains how to apply magnification to an entire document, over and above any magnification that has been specified when fonts are loaded. For example, if you have loaded a font that is scaled by |\magstep1| and if you also specify ^|\magnification||=\magstep2|, the actual font used for printing will be scaled by |\magstep3|. Similarly, if you load a font scaled by |\magstephalf| and if you also say |\magnification=\magstephalf|, the printed results will be scaled by |\magstep1|. \endchapter Type faces---like people's faces---have distinctive features indicating aspects of character. % I don't think he was kidding \author MARSHALL ^{LEE}, {\sl Bookmaking\/} (1965) % page 83 \bigskip This was the Noblest Roman of them all. \author WILLIAM ^{SHAKESPEARE}, {\sl The Tragedie % of Julius C\ae sar\/} (1599) % Act V, Scene 5, line 68 % For Shakespeare I'm using the spelling from First Folio (1623) % (titles not from the contents page, but the running heads in the plays) % but act/line numbers from The Riverside Shakespeare (throughout) \eject \beginchapter Chapter 5. Grouping Every once in a while it is necessary to treat part of a manuscript as a unit, so you need to indicate somehow where that part begins and where it ends. For this purpose \TeX\ gives special interpretation to two ``^{grouping characters},'' which (like the escape character) are treated differently from the normal symbols that you type. We assume in this manual that |{| and |}| are the grouping characters, since they are the ones used in plain \TeX. ^^{curly braces, see braces} We saw examples of grouping in the previous chapter, where it was mentioned that font changes inside a group do not affect the fonts in force outside. The same principle applies to almost anything else that is defined inside a group, as we will see later; for example, if you define a control sequence within some group, that definition will disappear when the group ends. In this way you can conveniently instruct \TeX\ to do something unusual, by changing its normal conventions temporarily inside of a group; since the changes are invisible from outside the group, there is no need to worry about messing up the rest of a manuscript by forgetting to restore the normal conventions when the unusual construction has been finished. Computer scientists have a name for this aspect of grouping, because it's an important aspect of programming languages in general; they call it ``^{block structure},'' and definitions that are in force only within a group are said to be ``^{local}'' to that group. You might want to use grouping even when you don't care about block structure, just to have better control over spacing. For example, let's consider once more the control sequence ^|\TeX| that produces the logo `\TeX' in this manual: We observed in Chapter~3 that a blank space after this control sequence will be gobbled up unless one types `\hbox{|\TeX\ |}', yet it is a mistake to say `|\TeX\|' when the following character is not a blank space. Well, in {\sl all\/} cases it would be correct to specify the simple group \begintt {\TeX} \endtt whether or not the following character is a ^{space}, because the |}| stops \TeX\ from absorbing an optional space into |\TeX|. This might come in handy when you're using a text editor (e.g., when replacing all occurrences of a particular word by a control sequence). Another thing you could do is type \begintt \TeX{} \endtt using an {\sl empty\/} group for the same purpose: The `|{}|' here is a group of no characters, so it produces no output, but it does have the effect of stopping \TeX\ from skipping blanks. ^^{empty group} ^^{lbrace rbrace} \exercise Sometimes you run into a rare word like `shelfful' that looks better as `shelf{}ful' without the `ff' ^{ligature}. How can you fool \TeX\ into thinking that there aren't two consecutive f's in such a word? \answer |{shelf}ful| or |shelf{}ful|, etc.; or even |shelf\/ful|, which yields a shelf\/ful instead of a shelf{\kern0pt}ful. In fact, the latter idea---to insert an ^{italic correction}---is preferable because \TeX\ will ^^|\/| reinsert the ff ligature by itself after ^{hyphenating} |shelf{}ful|. \ (Appendix~H points out that ligatures are put into a hyphenated word that contains no ``^{explicit kerns},'' and an italic correction is an explicit kern.) \ But the italic correction may be too much (especially in an italic font); |shelf{|^|\kern||0pt}ful| is often best. \dangerexercise Explain how to get three blank spaces in a row without using `|\|\]'.^^{control space} \answer `\]|{|\]|}|\]' or `\]|{}|\]|{}|\]', etc. Plain \TeX\ also has a ^|\space| macro, so you can type |\space\space\space|. \ (These aren't strictly equivalent to `|\|\]|\|\]|\|\]', since they adjust the spaces by the current ``^{space factor},'' as explained later.) But \TeX\ also uses grouping for another, quite different, purpose, namely to determine how much of your text is to be governed by certain control sequences. For example, if you want to center something on a line you can type \begintt \centerline{This information should be centered.} \endtt using the control sequence ^|\centerline| defined in plain \TeX\ format. Grouping is used in quite a few of \TeX's more intricate instructions; and it's possible to have groups within groups within groups, as you can see by glancing at Appendix~B\null. Complex grouping is generally unnecessary, however, in ordinary manuscripts, so you needn't worry about it. Just don't forget to finish each group that you've started, because a lost `|}|' might cause trouble. Here's an example of two groups, one ^{nested} inside the other: \begintt \centerline{This information should be {\it centered}.} \endtt As you might expect, \TeX\ will produce a centered line that also contains italics: $$\hbox{This information should be {\it centered}.}$$ But let's look at the example more closely: `|\centerline|' appears outside the curly braces, while `|\it|' appears inside. Why are the two cases different? And how can a beginner learn to remember which is which? Answer: |\centerline| is a control sequence that applies only to the very next thing that follows, so you want to put braces around the text that is to be centered (unless that text consists of a single symbol or control sequence). For example, to center the \TeX\ logo on a line, it would suffice to type `|\centerline\TeX|', but to center the phrase `\TeX\ has groups' you need braces: `|\centerline{\TeX\ has groups}|'. On the other hand, |\it| is a control sequence that simply means ``change the current font''; it acts without looking ahead, so it affects {\sl everything\/} that follows, at least potentially. The braces surround |\it| in order to confine the font change to a local region. In other words, the two sets of braces in this example actually have different functions: One serves to treat several words of the text as if they were a single object, while the other provides local block structure. \exercise What do you think happens if you type the following: \begintt \centerline{This information should be {centered}.} \centerline So should this. \endtt \answer In the first case, you get the same result as if the innermost braces had not appeared at all, because you haven't used the grouping to change fonts or to control spacing or anything. \TeX\ doesn't mind if you want to waste your time making groups for no particular reason. But in the second case, the necessary braces were forgotten. You get the letter `S' centered on a line by itself, followed by a paragraph that begins with `o should this.' on the next line. \exercise And how about this one? \begintt \centerline{This information should be \it centered.} \endtt \answer You get the same result as if another pair of braces were present around `|\it centered|', except that the period is typeset from the italic font. \ (Both periods look about the same.) \ The |\it| font will not remain in force after the |\centerline|, but this is something of a coincidence: \TeX\ uses the braces to determine what text is to be centered, but then it removes the braces. The |\centerline| operation, as defined in Appendix~B\null, puts the resulting braceless text inside {\sl another\/} group; and that's why |\it| disappears after |\centerline|. \ (If you don't understand this, just don't risk leaving out braces in tricky situations, and you'll be OK.) \smallskip \dangerexercise Define a control sequence |\ital| so that a user could type `|\ital{text}|' instead of `|{\it text\/}|'. Discuss the pros and cons of |\ital| versus |\it|. \answer |\def\ital#1{{\it#1\/}}|. \ Pro:~Users might find this easier to learn, because it works more like |\centerline| and they don't have to remember to make the italic correction. \ Con:~To avoid the italic correction just before a {\it comma} or {\it period}, users should probably be taught another control sequence; for example, with \begintt \def\nocorr{\kern0pt } \endtt a user could type `|\ital{comma} or \ital{period\nocorr},|'. The alternative of putting a period or comma in italics, to avoid the italic correction, doesn't look as good. A long sequence of italics would be inefficient for \TeX, since the entire text for the argument to |\ital| must be read into memory only to be scanned again. \ddanger Subsequent chapters describe many primitive operations of \TeX\ for which the locality of grouping is important. For example, if one of \TeX's internal parameters is changed within a group, the previous contents of that parameter will be restored when the group ends. Sometimes, however, it's desirable to make a definition that transcends its current group. This effect can be obtained by prefixing `^|\global|' to the definition. For example, \TeX\ keeps the current page number in a register called~|\count0|, and the routine that outputs a page wants to increase the ^{page number}. ^{Output routines} are always protected by enclosing them in groups, so that they do not inadvertently mess up the rest of \TeX; but the change to |\count0| would disappear if it were kept local to the output group. The command \begintt \global\advance\count0 by 1 \endtt ^^|\advance| solves the problem; it increases |\count0| and makes this value stick around at the end of the output routine. In general, |\global| makes the immediately following definition pertain to all existing groups, not just to the innermost one. \ddangerexercise If you think you understand local and global definitions, here's a little test to make sure: Suppose |\c| stands for `|\count1=|', |\g| stands for `|\global\count1=|', and |\s| stands for `|\showthe\count1|'. What values will be shown? \begintt {\c1\s\g2{\s\c3\s\g4\s\c5\s}\s\c6\s}\s \endtt \answer |{1 {2 3 4 5} 4 6} 4|. \ddanger Another way to obtain block structure with \TeX\ is to use the primitives ^|\begingroup| and ^|\endgroup|. These control sequences make it easy to begin a group within one control sequence and end it within another. The text that \TeX\ actually executes, after control sequences have been expanded, must have properly ^{nested groups}, i.e., groups that don't overlap. For example, \begintt { \begingroup } \endgroup \endtt is not legitimate. \ddangerexercise Define control sequences |\beginthe|\ and |\endthe|\ that provide a ``named'' block structure. In other words, \begintt \beginthe{beguine}\beginthe{waltz}\endthe{waltz}\endthe{beguine} \endtt should be permissible, but not \begintt \beginthe{beguine}\beginthe{waltz}\endthe{beguine}\endthe{waltz}. \endtt \answer |\def\beginthe#1{\begingroup\def\blockname{#1}}|\parbreak |\def\endthe#1{\def\test{#1}%|\parbreak | \ifx\test\blockname\endgroup|\parbreak | \else\errmessage{You should have said|\parbreak | \string\endthe{\blockname}}\fi}| \endchapter I have had recourse to varieties of type, and to braces. \author JAMES ^{MUIRHEAD}, {\sl The Institutes of Gaius\/} (1880) % p. xii % he actually said braces for what we call brackets \bigskip An encounter group is a gathering, for a few hours or a few days, of twelve or eighteen personable, responsible, certifiably normal and temporarily smelly people. \author JANE ^{HOWARD}, {\sl Please Touch\/} (1970) \eject \beginchapter Chapter 6. Running\\\TeX The best way to learn how to use \TeX\ is to use it. Thus, it's high time for you to sit down at a computer terminal and interact with the \TeX\ system, trying things out to see what happens. Here are some small but complete examples suggested for your first encounter. ^^{Running the program} Caution: This chapter is rather a long one. Why don't you stop reading now, and come back fresh tomorrow? \smallskip OK, let's suppose that you're rested and excited about having a trial run of \TeX\null. Step-by-step instructions for using it appear in this chapter. First do this: Go to the lab where the graphic output device is, since you will be wanting to see the output that you get---it won't really be satisfactory to run \TeX\ from a remote location, where you can't hold the generated documents in your own hands. Then log in; and start \TeX. \ (You may have to ask somebody how to do this on your local computer. Usually the operating system prompts you for a command and you type `|tex|' or `|run| |tex|' or something like that.) When you're successful, \TeX\ will welcome you with a message such as \begintt This is TeX, Version 3.141 (preloaded format=plain 89.7.15) ** \endtt The `^|**|' is \TeX's way of asking you for an input file name. % Incidentally, 89.7.15 was Jill's 50th birthday. Now type `^|\relax|' (including the backslash), and ^\ (or whatever is used to mean ``end-of-line'' on your terminal). \TeX\ is all geared up for action, ready to read a long manuscript; but you're saying that it's all right to take things easy, since this is going to be a real simple run. In fact, |\relax| is a control sequence that means ``do nothing.'' The machine will type another asterisk at you. This time type something like `|Hello?|'\ and wait for another ^{asterisk}. Finally type `^|\end|', and stand back to see what happens. \TeX\ should respond with `^|[1]|' (meaning that it has finished page~1 of your output); then the program will halt, probably with some indication that it has created a file called `|texput.dvi|'. \ (\TeX\ uses the name ^|texput| for its output when you haven't specified any better name in your first line of input; and ^|dvi| stands for ``^{device independent},'' since |texput.dvi| is capable of being printed on almost any kind of typographic output device.) Now you're going to need some help again from your friendly local computer hackers. They will tell you how to produce hardcopy from |texput.dvi|. And when you see the hardcopy---Oh, glorious day!---you will see a magnificent `Hello?'\ and the page number `1' at the bottom. Congratulations on your first masterpiece of fine printing. \smallbreak The point is, you understand now how to get something through the whole cycle. It only remains to do the same thing with a somewhat longer document. So our next experiment will be to work from a file instead of typing the input online. Use your favorite text editor to create a file called ^|story.tex| that contains the following 18 lines of text (no more, no less): $$\halign{\hbox to\parindent{\hfil\sevenrm#\ \ }&#\hfil\cr 1&|\hrule|\cr\noalign{^^|\hrule|} 2&|\vskip 1in|\cr\noalign{^^|\vskip|^^{leading, see vskip}} 3&|\centerline{\bf A SHORT STORY}|\cr\noalign{^^|\centerline|} 4&|\vskip 6pt|\cr 5&|\centerline{\sl by A. U. Thor}|\cr\noalign{^^{Thor}} 6&|\vskip .5cm|\cr 7&|Once upon a time, in a distant|\cr 8&| galaxy called \"O\"o\c c,|\cr\noalign{^^|\"|^^|\c|} 9&|there lived a computer|\cr 10&|named R.~J. Drofnats.|\cr\noalign{^^{Drofnats}} 11&||\cr 12&|Mr.~Drofnats---or ``R. J.,'' as|\cr 13&|he preferred to be called---|\cr 14&|was happiest when he was at work|\cr 15&|typesetting beautiful documents.|\cr 16&|\vskip 1in|\cr 17&|\hrule|\cr 18&|\vfill\eject|\cr\noalign{^^|\vfill|^^|\eject|}}$$ \write16{\ifnum\pageno=\storypage \else Redefine \string\storypage to be \the\pageno\fi}% (Don't type the numbers at the left of these lines, of course; they are present only for reference.) \ This example is a bit long, and more than a bit silly; but it's no trick for a good typist like you and it will give you some worthwhile experience, so do it. For your own good. And think about what you're typing, as you go; the example introduces a few important features of \TeX\ that you can learn as you're making the file. Here is a brief explanation of what you have just typed: Lines 1 and~17 put a horizontal ^{rule} (a thin line) across the page. Lines 2 and~16 skip past one inch of space; `|\vskip|' means ``vertical skip,'' and this extra space will separate the horizontal rules from the rest of the copy. Lines 3 and~5 produce the title and the author name, centered, in boldface and in slanted type. Lines 4 and~6 put extra white space between those lines and their successors. \ (We shall discuss units of measure like `|6pt|' and `|.5cm|' in Chapter~10.) The main bulk of the story appears on lines 7--15, and it consists of two ^{paragraphs}. The fact that line~11 is blank informs \TeX\ that ^^{blank line} ^^{empty line} line~10 is the end of the first paragraph; and the `|\vskip|' on line~16 implies that the second paragraph ends on line~15, because vertical skips don't appear in paragraphs. Incidentally, this example seems to be quite full of \TeX\ commands; but it is atypical in that respect, because it is so short and because it is supposed to be teaching things. Messy constructions like |\vskip| and |\centerline| can be expected at the very beginning of a manuscript, unless you're using a canned format, but they don't last long; most of the time you will find yourself typing straight text, with relatively few control sequences. And now comes the good news, if you haven't used computer typesetting before: You don't have to worry about where to break lines in a paragraph (i.e., where to stop at the right margin and to begin a new line), because \TeX\ will do that for you. Your manuscript file can contain long lines or short lines, or both; it doesn't matter. This is especially helpful when you make changes, since you don't have to retype anything except the words that changed. {\sl Every time you begin a new line in your manuscript file it is essentially the same as typing a space.} When \TeX\ has read an entire paragraph---in this case lines 7 to~11---it will try to break up the text so that each line of output, except the last, contains about the same amount of copy; and it will hyphenate words if necessary to keep the spacing consistent, but only as a last resort. Line 8 contains the strange concoction \begintt \"O\"o\c c \endtt and you already know that |\"| stands for an ^{umlaut} accent. The |\c| stands for a ``^{cedilla},'' so you will get `\"O\"o\c c' as the name of that distant galaxy. The remaining text is simply a review of the conventions that we discussed long ago for dashes and quotation marks, except that the `|~|' signs in lines 10 and~12 are a new wrinkle. These are called {\sl ^{ties}}, because they tie words together; i.e., \TeX\ is supposed to treat `|~|' as a normal space but not to break between lines there. A good typist will use ties within names, as shown in our example; further discussion of ties appears in Chapter~14. ^^{tilde} Finally, line~18 tells \TeX\ to `^|\vfill|', i.e., to fill the rest of the page with white space; and to `^|\eject|' the page, i.e., to send it to the output file. \smallskip Now you're ready for Experiment~2: Get \TeX\ going again. This time when the machine says `|**|' you should answer `|story|', since that is the name of the file where your input resides. \ (The file could also be called by its full name `|story.tex|', but \TeX\ automatically supplies the suffix `|.tex|' if no suffix has been specified.) ^^{file names} You might wonder why the first prompt was `^|**|', while the subsequent ones are `^|*|'; the reason is simply that the first thing you type to \TeX\ is slightly different from the rest: If the first character of your response to `|**|' is not a backslash, \TeX\ automatically inserts `^|\input|'. Thus you can usually run \TeX\ by merely naming your input file. \ (Previous \TeX\ systems required you to start by typing `|\input story|' instead of `|story|', and you can still do that; but most \TeX\ users prefer to put all of their commands into a file instead of typing them online, so \TeX\ now spares them the nuisance of starting out with |\input| each time.) \ Recall that in Experiment~1 you typed `|\relax|'; that started with a backslash, so |\input| was not implied. \danger There's actually another difference between `|**|' and `|*|': If the first character after |**| is an ^{ampersand} (\thinspace`|&|'\thinspace), \TeX\ will replace its memory with a precomputed ^{format file} before proceeding. Thus, for example, you can type `|&plain \input story|' or even `|&plain story|' in response to `|**|', if you are running some version of \TeX\ that might not have the plain format preloaded. ^^{preloaded formats} \danger Incidentally, many systems allow you to invoke \TeX\ by typing a one-liner like `|tex story|' instead of waiting for the `|**|'; similarly, `|tex \relax|' works for Experiment~1, and `|tex &plain story|' loads the plain format before inputting the |story| file. You might want to try this, to see if it works on your computer, or you might ask somebody if there's a similar shortcut. As \TeX\ begins to read your story file, it types `|(story.tex|', possibly with a version number for more precise identification, depending on your local operating system. Then it types `|[1]|', meaning that page~1 is done; and `|)|', meaning that the file has been entirely input. \TeX\ will now prompt you with `|*|', because the file did not contain `^|\end|'. Enter |\end| into the computer now, and you should get a file |story.dvi| containing a typeset version of Thor's story. As in Experiment~1, you can proceed to convert |story.dvi| into hardcopy; go ahead and do that now. The typeset output won't be shown here, but you can see the results by doing the experiment personally. Please do so before reading on. \exercise Statistics show that only 7.43 of 10 people who read this manual actually type the |story.tex| file as recommended, but that those people learn \TeX\ best. So why don't you join them? \answer Laziness and/or obstinacy. \exercise Look closely at the output of Experiment~2, and compare it to |story.tex|\thinspace: If you followed the instructions carefully, you will notice a typographical error. What is it, and why did it sneak in? \answer There's an unwanted space after `called---', because (as the book says) \TeX\ treats the end of a line as if it were a blank space. That blank space is usually what you want, except when a line ends with a hyphen or a dash; so you should {\sc WATCH OUT} for lines that end with hyphens or dashes. With Experiment 2 under your belt, you know how to make a document from a file. The remaining experiments in this chapter are intended to help you cope with the inevitable anomalies that you will run into later; we will intentionally do things that will cause \TeX\ to ``squeak.'' But before going on, it's best to fix the error revealed by the previous output (see exercise 6.2): Line~13 of the |story.tex| file should be changed to \begintt he preferred to be called---% error has been fixed! \endtt The `|%|' sign here ^^{percent} is a feature of plain \TeX\ that we haven't discussed before: It effectively terminates a line of your input file, without introducing the blank space that \TeX\ ordinarily inserts when moving to the next line of input. Furthermore, \TeX\ ignores everything that you type following a |%|, up to the end of that line in the file; you can therefore put ^{comments} into your manuscript, knowing that the comments are for your eyes only. Experiment 3 will be to make \TeX\ work harder, by asking it to set the story in narrower and narrower columns. Here's how: After starting the program, type \begintt \hsize=4in \input story \endtt in response to the `|**|'. This means, ``Set the story in a 4-inch column.'' More precisely, ^|\hsize| is a primitive of \TeX\ that specifies the horizontal size, i.e., the width of each normal line in the output when a paragraph is being typeset; and ^|\input| is a primitive that causes \TeX\ to read the specified file. Thus, you are instructing the machine to change the normal setting of\/ |\hsize| that was defined by plain \TeX, and then to process |story.tex| under this modification. \TeX\ should respond by typing something like `|(story.tex [1])|' as before, followed by `|*|'. Now you should type \begintt \hsize=3in \input story \endtt and, after \TeX\ says `|(story.tex [2])|' asking for more, type three more lines \begintt \hsize=2.5in \input story \hsize=2in \input story \end \endtt to complete this four-page experiment. Don't be alarmed when \TeX\ screams `|Overfull| |\hbox|' several times as it works at the 2-inch size; that's what was supposed to go wrong during Experiment~3. There simply is no good way to break the given paragraphs into lines that are exactly two inches wide, without making the spaces between words come out too large or too small. Plain \TeX\ has been set up to ensure rather strict tolerances on all of the lines it produces: \begindisplay \hbox spread-1em{You don't get spaces between words narrower than this,\ and}\cr \hbox spread+1.679895em{you don't get spaces between words wider than this.}\cr \enddisplay If there's no way to meet these restrictions, you get an ^{overfull box}. And with the overfull box you also get (1)~a warning message, printed on your terminal, and (2)~a big black bar inserted at the right of the offending box, in your output. \ (Look at page~4 of the output from Experiment~3; the overfull boxes should stick out like sore thumbs. On the other hand, pages 1--3 should be perfect.) Of course you don't want overfull boxes in your output, so \TeX\ provides several ways to remove them; that will be the subject of our Experiment~4. But first let's look more closely at the results of Experiment~3, since \TeX\ reported some potentially valuable information when it was forced to make those boxes too full; you should learn how to read this data: \begintt Overfull \hbox (0.98807pt too wide) in paragraph at lines 7--11 \tenrm tant galaxy called []O^^?o^^Xc, there lived|| Overfull \hbox (0.4325pt too wide) in paragraph at lines 7--11 \tenrm a com-puter named R. J. Drof-nats. || Overfull \hbox (5.32132pt too wide) in paragraph at lines 12--16 \tenrm he pre-ferred to be called---was hap-|| \endtt Each overfull box is correlated with its location in your input file (e.g., the first two were generated when processing the paragraph on lines 7--11 of |story.tex|), and you also learn by how much the copy sticks out (e.g., 0.98807 points). Notice that \TeX\ also shows the contents of the overfull boxes in abbreviated form. For example, the last one has the words `he preferred to be called---was hap-', set in font |\tenrm| (10-point roman type); the first one has a somewhat curious rendering of `\"O\"o\c c', because the accents appear in strange places within that font. In general, when you see `^|[]|' in one of these messages, it stands either for the paragraph indentation or for some sort of complex construction; in this particular case it stands for an umlaut that has been raised up to cover an `O'. \dangerexercise Can you explain the `\|' that appears after `|lived|' in that message? \answer It represents the heavy bar that shows up in your output. \ (This bar wouldn't be present if\/ ^|\overfullrule| had been set to |0pt|, nor is it present in an underfull box.) \ddangerexercise Why is there a space before the `\|' in `|Drof-nats. |\|'\thinspace? \answer This is the ^|\parfillskip| space that ends the paragraph. In plain \TeX\ the parfillskip is zero when the last line of the paragraph is full; hence no space actually appears before the rule in the output of Experiment~3. But all hskips show up as spaces in an overfull box message, even if they're zero. You don't have to take out pencil and paper in order to write down the overfull box messages that you get before they disappear from view, since \TeX\ always writes a ``^{transcript}'' or ``^{log file}'' that records what happened during each session. For example, you should now have a file called |story.log| containing the transcript of Experiment~3, as well as a file called |texput.log| containing the transcript of Experiment~1. \ (The transcript of Experiment~2 was probably overwritten when you did number~3.) \ Take a look at |story.log| now; you will see that the overfull box messages are accompanied not only by the abbreviated box contents, but also by some strange-looking data about hboxes and glue and kerns and such things. This data gives a precise description of what's in that overfull box; \TeX\ wizards will find such listings important, if they are called upon to diagnose some mysterious error, and you too may want to understand \TeX's internal code some day. The abbreviated forms of overfull boxes show the hyphenations that \TeX\ tried before it resorted to overfilling. The ^{hyphenation} algorithm, which is described in Appendix~H\null, is excellent but not perfect; for example, you can see from the messages in |story.log| that \TeX\ finds the hyphen in `pre-ferred', and it can even hyphenate `Drof-nats'. Yet it discovers no hyphen in `galaxy', and every once in a~while an overfull box problem can be cured simply by giving \TeX\ a hint about how to hyphenate some word more completely. \ (We will see later that there are two ways to do this, either by inserting ^{discretionary hyphens} each time as in `\hbox{|gal\-axy|}', or by saying `\hbox{|\hyphenation{gal-axy}|}' once at the beginning of your manuscript.) In the present example, hyphenation is not a problem, since \TeX\ found and tried all the hyphens that could possibly have helped. The only way to get rid of the overfull boxes is to change the tolerance, i.e., to allow wider spaces between words. Indeed, the tolerance that plain \TeX\ uses for wide lines is completely inappropriate for 2-inch columns; such narrow columns simply can't be achieved without loosening the constraints, unless you rewrite the copy to fit. \TeX\ assigns a numerical value called ``^{badness}'' to each line that it sets, in order to assess the quality of the spacing. The exact rules for badness are different for different fonts, and they will be discussed in Chapter~14; but here is the way badness works for the roman font of plain \TeX: \begindisplay \hbadness10000 \hbox spread-.666667em{The badness of this line is 100.}& \quad(very tight)\cr \hbox spread-.333333em{The badness of this line is 12.}& \quad(somewhat tight)\cr \hbox{The badness of this line is 0.}& \quad(perfect)\cr \hbox spread.5em{The badness of this line is 12.}& \quad(somewhat loose)\cr %\hbox spread 1em{The badness of this line is 100.}& % \quad(loose)\cr % then "looser" \hbox spread 1.259921em{The badness of this line is 200.}& \quad(loose)\cr %\hbox spread 1.713em{The badness of this line is 500.}& % \quad(bad)\cr % then "worse" \hbox spread 2.155em{The badness of this line is 1000.}& \quad(bad)\cr \hbox spread 3.684em{The badness of this line is 5000.}& % actually 4995! \quad(awful)\cr \enddisplay Plain \TeX\ normally stipulates that no line's badness should exceed 200; but in our case, the task would be impossible since \begindisplay \hbadness 10000 `\hbox to 2in{tant galaxy called \"O\"o\c c, there}'\hskip 3em has badness 1521;\cr `\hbox to 2in{he preferred to be called---was}'\hskip 3em has badness 568.\cr \enddisplay So we turn now to Experiment~4, in which spacing variations that are more appropriate to narrow columns will be used. Run \TeX\ again, and begin this time by saying \begintt \hsize=2in \tolerance=1600 \input story \endtt so that lines with badness up to 1600 will be tolerated. Hurray! There are ^^|\tolerance| no overfull boxes this time. \ (But you do get a message about an {\sl underfull\/} box, since \TeX\ reports all boxes whose badness exceeds a certain threshold called ^|\hbadness|; plain \TeX\ sets |\hbadness=1000|.) \ ^^{underfull box} Now make \TeX\ work still harder by trying \begintt \hsize=1.5in \input story \endtt (thus leaving the tolerance at 1600 but making the ^{column width} still ^^{measure, see hsize} skimpier). Alas, overfull boxes return; so try typing \begintt \tolerance=10000 \input story \endtt in order to see what happens. \TeX\ treats 10000 as if it were ``infinite'' tolerance, allowing arbitrarily wide space; thus, a tolerance of 10000 will {\sl never\/} produce an overfull box, unless something strange occurs like an unhyphenatable word that is wider than the column itself. The underfull box that \TeX\ produces in the 1.5-inch case is really bad; with such narrow limits, an occasional wide space is unavoidable. But try \begintt \raggedright \input story \endtt for a change. \ ^^|\raggedright|(This tells \TeX\ not to worry about keeping the right margin straight, and to keep the spacing uniform within each line.) \ Finally, type \begintt \hsize=.75in \input story \endtt followed by `|\end|', to complete Experiment 4. This makes the columns almost impossibly narrow. \danger The output from this experiment will give you some feeling for the problem of ^{breaking a paragraph} into approximately equal lines. When the lines are relatively wide, \TeX\ will almost always find a good solution. But otherwise you will have to figure out some compromise, and several options are possible. Suppose you want to ensure that no lines have badness exceeding~500. Then you could set |\tolerance| to some high number, and |\hbadness=500|; \TeX\ would not produce overfull boxes, but it would warn you about the underfull ones. Or you could set |\tolerance=500|; then \TeX\ might produce overfull boxes. If you really want to take corrective action, the second alternative is better, because you can look at an overfull box to see how much sticks out; it becomes graphically clear what remedies are possible. On the other hand, if you don't have time to fix bad spacing---if you just want to know how bad it is---then the first alternative is better, although it may require more computer time. \dangerexercise When |\raggedright| has been specified, badness reflects the amount of space at the right margin, instead of the spacing between words. Devise an experiment by which you can easily determine what badness \TeX\ assigns to each line, when the |story| is set ragged-right in 1.5-inch columns. \answer Run \TeX\ with \hbox{|\hsize=1.5in|} \hbox{|\tolerance=10000|} \hbox{|\raggedright|} \hbox{|\hbadness=-1|} and then |\input story|. \TeX\ will report the badness of all lines (except the final lines of paragraphs, where fill glue makes the badness zero). \danger A parameter called ^|\hfuzz| allows you to ignore boxes that are only slightly overfull. For example, if you say |\hfuzz=1pt|, a box must stick out more than one point before it is considered erroneous. Plain \TeX\ sets |\hfuzz=0.1pt|. \ddangerexercise Inspection of the output from Experiment~4, especially page~3, shows that with narrow columns it would be better to allow white space to appear before and after a dash, whenever other spaces in the same line are being stretched. Define a ^|\dash| macro that does this. \answer |\def\extraspace{\nobreak \hskip 0pt plus .15em\relax}|\parbreak |\def\dash{\unskip\extraspace---\extraspace}|\par\nobreak\smallskip\noindent (If you try this with the story at 2-inch and 1.5-inch sizes, you will notice a substantial improvement. The |\unskip| allows people to leave a space before typing |\dash|. \TeX\ will try to hyphenate before |\dash|, but not before `|---|'; cf.\ Appendix~H\null. The ^|\relax| at the end of |\extraspace| is a precaution in case the next word is `|minus|'.) You were warned that this is a long chapter. But take heart: There's only one more experiment to do, and then you will know enough about \TeX\ to run it fearlessly by yourself forever after. The only thing you are still missing is some information about how to cope with ^{error messages}---i.e., not just with warnings about things like overfull boxes, but with cases where \TeX\ actually stops and asks you what to do next. Error messages can be terrifying when you aren't prepared for them; but they can be fun when you have the right attitude. Just remember that you really haven't hurt the computer's feelings, and that nobody will hold the errors against you. Then you'll find that running \TeX\ might actually be a creative experience instead of something to dread. The first step in Experiment 5 is to plant two intentional mistakes in the |story.tex| file. Change line~3 to \begintt \centerline{\bf A SHORT \ERROR STORY} \endtt and change `|\vskip|' to `|\vship|' on line~2. Now run \TeX\ again; but instead of `|story|' type `|sorry|'. The computer should respond by saying that it can't find file |sorry.tex|, and it will ask you to try again. Just hit \ this time; you'll see that you had better give the name of a real file. So type `|story|' and wait for \TeX\ to find one of the {\sl faux pas\/} in that file. Ah yes, the machine will soon stop,\footnote*{Some installations of \TeX\ do not allow interaction. In such cases all you can do is look at the error messages in your log file, where they will appear together with the ``help'' information.} after typing something like this: \begintt ! Undefined control sequence. l.2 \vship 1in ? \endtt \write16{\ifnum\pageno=\vshippage \else Redefine \string\vshippage to be \the\pageno\fi}% \TeX\ begins its error messages with `|!|', and it shows what it was reading at the time of the error by displaying two lines of context. The top line of the pair (in this case `|\vship|'\thinspace) shows what \TeX\ has looked at so far, and where it came from (`|l.2|', i.e., line number~2); the bottom line (in this case `|1in|'\thinspace) shows what \TeX\ has yet to read. The `^|?|'\ that appears after the context display means that \TeX\ wants advice about what to do next. If you've never seen an error message before, or if you've forgotten what sort of response is expected, you can type `|?|'\ now (go ahead and try it!); \TeX\ will respond as follows: \begintt Type to proceed, S to scroll future error messages, R to run without stopping, Q to run quietly, I to insert something, E to edit your file, 1 or ... or 9 to ignore the next 1 to 9 tokens of input, H for help, X to quit. \endtt This is your menu of options. You may choose to continue in various ways: \smallskip\item{1.} Simply type \. \TeX\ will resume its processing, after attempting to recover from the error as best it can. \smallbreak\item{2.} Type `|S|'. \TeX\ will proceed without pausing for instructions if further errors arise. Subsequent error messages will flash by on your terminal, possibly faster than you can read them, and they will appear in your log file where you can scrutinize them at your leisure. Thus, `|S|' is sort of like typing \ to every message. \smallbreak\item{3.} Type `|R|'. This is like `|S|' but even stronger, since it tells \TeX\ not to stop for any reason, not even if a file name can't be found. \smallbreak\item{4.} Type `|Q|'. This is like `|R|' but even more so, since it tells \TeX\ not only to proceed without stopping but also to suppress all further output to your terminal. It is a fast, but somewhat reckless, way to proceed (intended for running \TeX\ with no operator in attendance). \smallbreak\item{5.} Type `|I|', followed by some text that you want to insert. \TeX\ will read this line of text before encountering what it would ordinarily see next. Lines inserted in this way are not assumed to end with a blank space. ^^{inserting text online} ^^{online interaction, see interaction} ^^{interacting with TeX} \smallbreak\item{6.} Type a small number (less than 100). \TeX\ will delete this many characters and control sequences from whatever it is about to read next, and it will pause again to give you another chance to look things over. ^^{deleting tokens} \smallbreak\item{7.} Type `|H|'. This is what you should do now and whenever you are faced with an error message that you haven't seen for a~while. \TeX\ has two messages built in for each perceived error: a formal one and an informal one. The formal message is printed first (e.g., `|! Undefined control sequence.|'\thinspace); the informal one is printed if you request more help by typing `|H|', and it also appears in your log file if you are scrolling error messages. The informal message tries to complement the formal one by explaining what \TeX\ thinks the trouble is, and often by suggesting a strategy for recouping your losses.^^{help messages} \smallbreak\item{8.} Type `|X|'. This stands for ``exit.'' It causes \TeX\ to stop working on your job, after putting the finishing touches on your |log| file and on any pages that have already been output to your |dvi| file. The current (incomplete) page will not be output. \smallbreak\item{9.} Type `|E|'. This is like `|X|', but it also prepares the computer to edit the file that \TeX\ is currently reading, at the current position, so that you can conveniently make a change before trying again. \smallbreak\noindent After you type `|H|' (or `|h|', which also works), you'll get a message that tries to explain that the control sequence just read by \TeX\ (i.e., |\vship|) has never been assigned a meaning, and that you should either insert the correct control sequence or you should go on as if the offending one had not appeared. In this case, therefore, your best bet is to type \begintt I\vskip \endtt (and \), with no space after the `|I|'; this effectively replaces |\vship| by |\vskip|. \ (Do it.) If you had simply typed \ instead of inserting anything, \TeX\ would have gone ahead and read `|1in|', which it would have regarded as part of a paragraph to be typeset. Alternatively, you could have typed `|3|'\thinspace; that would have deleted `|1in|' from \TeX's input. Or you could have typed `|X|' or `|E|' in order to correct the spelling error in your file. But it's usually best to try to detect as many errors as you can, each time you run \TeX, since that increases your productivity while decreasing your computer bills. Chapter~27 explains more about the art of steering \TeX\ through troubled text. \dangerexercise What would have happened if you had typed `|5|' after the |\vship| error? \answer \TeX\ would have deleted five tokens: |1|, |i|, |n|, \], |\centerline|. (The space was at the end of line~2, the |\centerline| at the beginning of line~3.) \danger You can control the level of interaction by giving commands in your file as well as online: The \TeX\ primitives ^|\scrollmode|, ^|\nonstopmode|, and ^|\batchmode| correspond respectively to typing `|S|', `|R|', or `|Q|' in response to an error message, and ^|\errorstopmode| puts you back into the normal level of interaction. \ (Such changes are global, whether or not they appear inside a group.) \ Furthermore, many installations have implemented a way to ^{interrupt} \TeX\ while it is running; such an interruption causes the program to revert to |\errorstopmode|, after which it pauses and waits for further instructions. What happens next in Experiment 5? \TeX\ will hiccup on the other bug that we planted in the file. This time, however, the error message is more elaborate, since the context appears on six lines instead of two: \begintt ! Undefined control sequence. \bf A SHORT \ERROR STORY \centerline #1->\line {\hss #1 \hss } l.3 \centerline{\bf A SHORT \ERROR STORY} |null ? \endtt You get multiline error messages like this when the error is detected while \TeX\ is processing some higher-level commands---in this case, while it is trying to carry out |\centerline|, which is not a primitive operation (it is defined in plain \TeX). At first, such error messages will appear to be complete nonsense to you, because much of what you see is low-level \TeX\ code that you never wrote. But you can overcome this hangup by getting a feeling for the way \TeX\ operates. First notice that the context information always appears in pairs of lines. As before, the top line shows what \TeX\ has just read (\thinspace `|\bf A SHORT \ERROR|'\thinspace), then comes what it is about to read (\thinspace`|STORY|'\thinspace). The next pair of lines shows the context of the first two; it indicates what \TeX\ was doing just before it began to read the others. In this case, we see that \TeX\ has just read `|#1|', which is a special code that tells the machine to ``read the first ^{argument} that is governed by the current control sequence''; i.e., ``now read the stuff that ^|\centerline| is supposed to center on a line.'' The definition in Appendix~B says that |\centerline|, when applied to some text, is supposed to be carried out by sticking that text in place of the `|#1|' in `|\line{\hss#1\hss}|'. So \TeX\ is in the midst of this expansion of\/ |\centerline|, as well as being in the midst of the text that is to be centered. \looseness-1 The bottom line shows how far \TeX\ has gotten until now in the |story| file. \ (Actually the bottom line is blank in this example; what appears to be the bottom line is really the first of two lines of context, and it indicates that \TeX\ has read everything including the `|}|' in line~3 of the file.) \ Thus, the context in this error message gives us a glimpse of how \TeX\ went about its business. First, it saw |\centerline| at the beginning of line~3. Then it looked at the definition of\/ |\centerline| and noticed that |\centerline| takes an ``argument,'' i.e., that |\centerline| applies to the next character or control sequence or group that follows. So \TeX\ read~on, and filed `|\bf A SHORT \ERROR STORY|' away as the argument to |\centerline|. Then it began to read the expansion, as defined in Appendix~B\null. When it reached the |#1|, it began to read the argument it had saved. And when it reached |\ERROR|, it complained about an undefined control sequence. \dangerexercise Why didn't \TeX\ complain about |\ERROR| being undefined when |\ERROR| was first encountered, i.e., before reading `|STORY}|' on line~3? \answer A control sequence like |\centerline| might well define a control sequence like |\ERROR| before telling \TeX\ to look at |#1|. Therefore \TeX\ doesn't interpret control sequences when it scans an argument. When you get a multiline error message like this, the best clues about the source of the trouble are usually on the bottom line (since that is what you typed) and on the top line (since that is what triggered the error message). Somewhere in there you can usually spot the problem. Where should you go from here? If you type `|H|' now, you'll just get the same help message about undefined control sequences that you saw before. If you respond by typing \, \TeX\ will go on and finish the run, producing output virtually identical to that in Experiment~2. In other words, the conventional responses won't teach you anything new. So type `|E|' now; this terminates the run and prepares the way for you to fix the erroneous file. \ (On some systems, \TeX\ will actually start up the standard text editor, and you'll be positioned at the right place to delete `|\ERROR|'. On other systems, \TeX\ will simply tell you to edit line~3 of file |story.tex|.) ^^{editing} When you edit |story.tex| again, you'll notice that line~2 still contains |\vship|; the fact that you told \TeX\ to insert |\vskip| doesn't mean that your file has changed in any way. In general, you should correct all errors in the input file that were spotted by \TeX\ during a run; the log file provides a handy way to remember what those errors were. \smallskip Well, this has indeed been a long chapter, so let's summarize what has been accomplished. By doing the five experiments you have learned at first hand (1)~how to get a job printed via \TeX; (2)~how to make a file that contains a complete \TeX\ manuscript; (3)~how to change the plain \TeX\ format to achieve columns with different widths; and (4)~how to avoid panic when \TeX\ issues stern~warnings. So you could now stop reading this book and go on to print a bunch of documents. It is better, however, to continue bearing with the author (after perhaps taking another rest), since you're just at the threshold of being able to do a lot more. And you ought to read Chapter~7 at least, because it warns you about certain symbols that you must not type unless you want \TeX\ to do something special. While reading the remaining chapters it will, of course, be best for you to continue making trial runs, using experiments of your own design. \ddanger If you use \TeX\ format packages designed by others, your error messages may involve many inscrutable two-line levels of macro context. By setting ^|\errorcontextlines||=0| at the beginning of your file, you can reduce the amount of information that is reported; \TeX\ will show only the top and bottom pairs of context lines together with up to |\errorcontextlines| additional two-line items. \ (If anything has thereby been omitted, you'll also see `|...|'.) \ Chances are good that you can spot the source of an error even when most of a large context has been suppressed; if not, you can say `|I\errorcontextlines=100\oops|' and try again. \ (That will usually give you an undefined control sequence error and plenty of context.) \ Plain \TeX\ sets |\errorcontextlines=5|. \endchapter What we have to learn to do we learn by doing. \author ^{ARISTOTLE}, {\sl Ethica Nicomachea\/} II (c.~325 B.C.) \bigskip He may run who reads. \author ^{HABAKKUK} 2\thinspace:\thinspace2 (c.~600 B.C.) % RSV \smallskip He that runs may read. \author WILLIAM ^{COWPER}, {\sl Tirocinium\/} (1785) \eject \beginchapter Chapter 7. How \TeX\ Reads\\What You Type We observed in the previous chapter that an input manuscript is expressed in terms of ``lines,'' but that these lines of input are essentially independent of the lines of output that will appear on the finished pages. Thus you can stop typing a line of input at any place that's convenient for you, as you prepare or edit a file. A few other related rules have also been mentioned: \medskip \item\bull A $\langle\hbox{return}\rangle$ is like a space. \smallskip \item\bull Two spaces in a row count as one space. \smallskip \item\bull A blank line denotes the end of a paragraph. \medskip \noindent Strictly speaking, these rules are contradictory: A blank line is obtained by typing $\langle\hbox{return}\rangle$ twice in a row, and this is different from typing two spaces in a row. Some day you might want to know the {\sl real\/} rules. In this chapter and the next, we shall study the very first stage in the transition from input to output. \smallskip In the first place, it's wise to have a precise idea of what your keyboard sends to the machine. There are 256 characters that \TeX\ might encounter at each step, in a file or in a line of text typed directly on your terminal. These 256~characters are classified into 16 categories numbered 0 to 15: \begindisplay \def\\{\hfill} \hfil\hidewidth\it Category\hidewidth&\it \qquad Meaning\hidewidth\cr \noalign{\smallskip} \\0&Escape character&(|\| in this manual)\cr \\1&Beginning of group&(|{| in this manual)\cr \\2&End of group&(|}| in this manual)\cr \\3&Math shift&(|$| in this manual)\cr \\4&Alignment tab&(|&| in this manual)\cr \\5&End of line&(\ in this manual)\cr \\6&Parameter&(|#| in this manual)\cr \\7&Superscript&(|^| in this manual)\cr \\8&Subscript&(|_| in this manual)\cr \\9&Ignored character&(\ in this manual)\cr 10&Space&(\] in this manual)\cr 11&Letter&(|A|, \dots, |Z| and |a|, \dots, |z|)\cr 12&Other character&(none of the above or below)\cr 13&Active character&(|~| in this manual)\cr 14&Comment character&(|%| in this manual)\cr 15&Invalid character&(\ in this manual)\cr \enddisplay ^^{escape character} ^^{begin-group character} ^^{end-group character} ^^{math mode character} ^^{alignment tab} ^^{parameter} ^^{superscript} ^^{subscript} ^^{ignored character} ^^{space} ^^{letter} ^^{other character} ^^{active character} ^^{comment character} ^^{invalid character} ^^{category codes, table} It's not necessary for you to learn these code numbers; the point is only that \TeX\ responds to 16~different types of characters. At first this manual led you to believe that there were just two types---the escape character and the others---and then you were told about two more types, the grouping symbols |{| and~|}|. In Chapter~6 you learned two more: |~| and~|%|. Now you know that there are really~16. This is the whole truth of the matter; no more types remain to be revealed. The category code for any character can be changed at any time, but it is usually wise to stick to a ^^{reserved character} ^^{special character table} ^^\ ^^\ particular scheme. The main thing to bear in mind is that each \TeX\ format reserves certain characters for its own special purposes. For example, when you are using plain \TeX\ format (Appendix~B\null), you need to know that the ten characters \begintt \ { } $ & # ^ _ % ~ \endtt cannot be used in the ordinary way when you are typing; ^^{special characters} ^^{backslash}^^{left brace}^^{right brace}^^{dollar sign}^^{ampersand} ^^{hash mark}^^{hat}^^{underline}^^{percent}^^{tilde} ^^{single-character control sequences} each of them will cause \TeX\ to do something special, as explained elsewhere in this book. If you really need these symbols as part of your manuscript, plain \TeX\ makes it possible for you to type \begindisplay |\$| for \$,\qquad |\%| for \%,\qquad |\&| for \&,\qquad |\#| for \#,\qquad |\_| for \_\thinspace; \enddisplay the |\_| symbol is useful for {\it compound\_identifiers\/} in computer ^^{identifiers} ^^{computer programs} programs. In mathematics formulas you can use |\{| and |\}| for $\{$ and $\}$, while ^|\backslash| produces a ^{reverse slash}; for example, \begindisplay `|$\{a \backslash b\}$|'\quad yields\quad`$\{a\backslash b\}$'. \enddisplay Furthermore |\^| produces a circumflex accent (e.g., `|\^e|' yields `\^e'\thinspace); and |\~| yields a tilde accent (e.g., `|\~n|' yields `\~n'\thinspace). \exercise What horrible errors appear in the following sentence? ^^{Procter} ^^{Gamble} \begintt Procter & Gamble's stock climbed to $2, a 10% gain. \endtt \answer Three forbidden characters were used. One should type \begintt Procter \& Gamble's ... \$2, a 10\% gain. \endtt (Also the facts are wrong.) \exercise Can you imagine why the designer of plain \TeX\ decided not to make `|\\|' the control sequence for reverse slashes?^^{backslash} \answer Reverse slashes (backslashes) are fairly uncommon in formulas or text, and |\\| is very easy to type; it was therefore felt best not to reserve |\\| for such limited use. Typists can define |\\| to be whatever they want (including |\backslash|). \danger When \TeX\ reads a line of text from a file, or a line of text that you entered directly on your terminal, it converts that text into a list of ``^{tokens}.'' A token is either (a)~a single character with an attached category code, or (b)~a control sequence. For example, if the normal conventions of plain \TeX\ are in force, the text `|{\hskip 36 pt}|' is converted into a list of eight tokens: \begindisplay |{|$_1$\quad\cstok{hskip}\quad|3|$_{12}$\quad|6|$_{12}$\quad \]$_{10}$\quad|p|$_{11}$\quad|t|$_{11}$\quad|}|$_{2}$ \enddisplay The subscripts here are the category codes, as listed earlier: 1 for ``beginning of group,'' 12 for ``other character,'' and so on. The \cstok{hskip} doesn't get a subscript, because it represents a control sequence token instead of a character token. Notice that the space after |\hskip| does not get into the token list, because it follows a ^{control word}. \danger It is important to understand the idea of token lists, if you want to gain a thorough understanding of \TeX, and it is convenient to learn the concept by thinking of \TeX\ as if it were a living organism. The individual lines of input in your files are seen only by \TeX's ``eyes'' and ``mouth''; but after that text has been gobbled up, it is sent to \TeX's ``stomach'' in the form of a token list, and the digestive processes that do the actual typesetting are based entirely on tokens. As far as the stomach is concerned, the input flows in as a stream of tokens, somewhat as if your \TeX\ manuscript had been typed all on one extremely long line. \danger You should remember two chief things about \TeX's tokens: (1)~A control sequence is considered to be a single object that is no longer composed of a sequence of symbols. Therefore long control sequence names are no harder for \TeX\ to deal with than short ones, after they have been replaced by tokens. Furthermore, spaces are not ignored after control sequences inside a token list; the ignore-space rule applies only in an input file, during the time that strings of characters are being tokenized. (2)~Once a category code has been attached to a character token, the attachment is permanent. For example, if character `|{|' were suddenly declared to be of category~12 instead of category~1, the characters `|{|$_1$' already inside token lists of \TeX\ would still remain of category 1; only newly made lists would contain `|{|$_{12}$' tokens. In other words, individual characters receive a fixed interpretation as soon as they have been read from a file, based on the category they have at the time of reading. Control sequences are different, since they can change their interpretation at any time. \TeX's digestive processes always know exactly what a character token signifies, because the category code appears in the token itself; but when the digestive processes encounter a control sequence token, they must look up the current definition of that control sequence in order to figure out what it means. \ddangerexercise Some of the category codes 0 to 15 will never appear as subscripts in character tokens, because they disappear in \TeX's mouth. For example, characters of category 0 (escapes) never get to be tokens. Which categories can actually reach \TeX's stomach? \answer 1, 2, 3, 4, 6, 7, 8, 10, 11, 12, 13. ^{Active characters} (type 13) are somewhat special; they behave like control sequences in most cases (e.g., when you say `^|\let||\x=~|' or `^|\ifx||\x~|'), but they behave like character tokens when they appear in the token list of\/ ^|\uppercase| or ^|\lowercase|, and when unexpanded after ^|\if| or ^|\ifcat|. \ddanger There's a program called ^|INITEX| that is used to install \TeX, starting from scratch; |INITEX| is like \TeX\ except that it can do even more things. It can compress ^{hyphenation} patterns into special tables that facilitate rapid hyphenation, and it can produce ^{format} files like `|plain.fmt|' from `|plain.tex|'. But |INITEX| needs extra space to carry out such tasks, so it generally has less memory available for typesetting than you would expect to find in a production version of \TeX. \ddanger When |INITEX| begins, it knows nothing but \TeX's primitives. All 256~characters are initially of category~12, except that ^\ has category~5, ^\ has category~10, ^\ has category~9, ^\ has category~15, the 52 letters |A|$\,\ldots\,$|Z| and |a|$\,\ldots\,$|z| have category~11, |%| and~|\| have the respective categories 14 and~0. ^^{backslash}^^{percent} It follows that |INITEX| is initially incapable of carrying out some of \TeX's primitives that depend on grouping; you can't use |\def| or |\hbox| until there are characters of categories 1 and~2. The format in Appendix~B begins with ^|\catcode| commands to provide characters of the necessary categories; e.g., \begintt \catcode`\{=1 \endtt assigns category 1 to the |{| symbol. The |\catcode| operation is like many other primitives of \TeX\ that we shall study later; by modifying internal quantities like the category codes, you can adapt \TeX\ to a wide variety of applications. \ddangerexercise Suppose that the commands \begintt \catcode`\<=1 \catcode`\>=2 \endtt appear near the beginning of a group that begins with `|{|'; these specifications instruct \TeX\ to treat |<| and |>| as group delimiters. According to \TeX's rules of locality, the characters |<| and |>| will revert to their previous categories when the ^{group} ends. But should the group end with |}| or~with~|>|\thinspace? \answer It ends with either |>| or |}| or any character of category 2; then the effects of all |\catcode| definitions within the group are wiped out, except those that were ^|\global|. \TeX\ doesn't have any built-in knowledge about how to pair up particular kinds of grouping characters. New category codes take effect as soon as a |\catcode| assignment has been digested. For example, \begintt {\catcode`\>=2 > \endtt is a complete group. But without the space after `|2|' it would not be complete, since \TeX\ would have read the~`|>|' and converted it to a token before knowing what category code was being specified; \TeX\ always reads the token following a constant before evaluating that ^{constant}. \ddanger Although control sequences are treated as single objects, \TeX\ does provide a way to break them into lists of character tokens: If you write ^|\string||\cs|, where |\cs| is any control sequence, you get the list of characters for that control sequence's name. For example, |\string\TeX| produces four tokens: |\|$_{12}$, |T|$_{12}$, |e|$_{12}$, |X|$_{12}$. Each character in this token list automatically gets category code~12 (``other''), including the ^{backslash} that |\string| inserts to represent an escape character. However, category~10 will be assigned to the character `\]' (blank ^{space}) if a space character somehow sneaks into the name of a control sequence. \ddanger Conversely, you can go from a list of character tokens to a control sequence by saying `^|\csname|\^|\endcsname|'. The tokens that appear in this construction between |\csname| and |\endcsname| may include other control sequences, as long as those control sequences ultimately expand into characters instead of \TeX\ primitives; the final characters can be of any category, not necessarily letters. For example, `|\csname TeX\endcsname|' is essentially the same as `|\TeX|'; but `|\csname\TeX\endcsname|' is illegal, because |\TeX| expands into tokens containing the ^|\kern| primitive. Furthermore, `|\csname\string\TeX\endcsname|' will produce the unusual control sequence `|\\TeX|', i.e., the token \cstok{\char`\\TeX}, which you can't ordinarily write. \ddangerexercise Experiment with \TeX\ to see what |\string| does when it is followed by an ^{active character} like |~|. \ (Active characters behave like control sequences, but they are not prefixed by an escape.) \ What is an easy way to conduct such experiments online? What control sequence could you put after |\string| to~obtain the single character token~|\|$_{12}$? \answer If you type `|\message{\string~}|' and `|\message{\string\~}|', \TeX\ responds with `|~|' and `|\~|', respectively. ^^|\message| To get |\|$_{12}$ from |\string| you therefore need to make backslash an active character. One way to do this is \begintt {\catcode`/=0 \catcode`\\=13 /message{/string\}} \endtt (The ``^{null control sequence}'' that you get when there are no tokens between |\csname| and |\endcsname| is not a solution to this exercise, because |\string| converts it to `|\csname\endcsname|'. There is, however, another solution: If \TeX's ^|\escapechar| parameter---which will be explained in one of the next dangerous bends---is negative or greater than~255, then `|\string\\|' works.) \ddangerexercise What tokens does `|\expandafter\string\csname a\string\ b\endcsname|' produce? (There are three spaces before the |b|. Chapter~20 explains ^|\expandafter|.) \answer |\|$_{12}$ |a|$_{12}$ |\|$_{12}$ \]$_{10}$ |b|$_{12}$. \ddangerexercise When |\csname| is used to define a control sequence for the first time, that control sequence is made equivalent to |\relax| until it is redefined. Use this fact to design a macro |\ifundefined#1| such that, for example, \begindisplay |\ifundefined{TeX}|\|\else|\|\fi| \enddisplay expands to the \ if\/ |\TeX| hasn't previously been defined, or if\/ |\TeX| has been |\let| equal to |\relax|; it should expand to the \ otherwise. ^^|\ifundefined| \answer |\def\ifundefined#1{\expandafter\ifx\csname#1\endcsname\relax}|% \hfil\break Note that a control sequence like this must be used with care; it cannot be included in ^{conditional} text, because the |\ifx| will not be seen when |\ifundefined| isn't expanded. \ddanger In the examples so far, |\string| has converted control sequences into lists of tokens that begin with |\|$_{12}$. But this backslash token isn't really hardwired into \TeX; there's a parameter called ^|\escapechar| that specifies what character should be used when control sequences are output as text. The value of\/ |\escapechar| is normally \TeX's internal code for backslash, but it can be changed if another convention is desired. \ddanger \TeX\ has two other token-producing operations similar to the |\string| command. If you write ^|\number|\, you get the decimal equivalent of the \; and if you write ^|\romannumeral|\, you get the number expressed in lowercase ^{roman numerals}. For example, `|\romannumeral24|' produces `|xxiv|', a list of four tokens each having category~12. The |\number| operation is redundant when it is applied to an explicit constant (e.g., `|\number24|' produces `|24|'); but it does suppress leading zeros, and it can also be used with numbers that are in \TeX's internal registers or parameters. For example, `|\number-0015|' produces `|-15|'; and if register |\count5| holds the value 316, then `|\number\count5|' produces `|316|'. \ddanger The twin operations ^|\uppercase||{|\|}| and ^|\lowercase||{|\|}| go through a given token list and convert all of the character tokens to their ``uppercase'' or ``lowercase'' equivalents. Here's how: Each of the 256 possible characters has two associated values called the ^|\uccode| and the ^|\lccode|; these values are changeable just as a |\catcode| is. Conversion to uppercase means that a character is replaced by its |\uccode| value, unless the |\uccode| value is zero (when no change is made). Conversion to lowercase is similar, using the |\lccode|. The category codes aren't changed. When ^|INITEX| begins, all |\uccode| and |\lccode| values are zero except that the ^{letters} |a| to~|z| and |A| to~|Z| have |\uccode| values |A| to~|Z| and |\lccode| values |a| to~|z|. \ddanger \TeX\ performs the |\uppercase| and |\lowercase| transformations in its stomach, but the |\string| and |\number| and |\romannumeral| and |\csname| operations are carried out en route to the stomach (like macro expansion), as explained in Chapter~20. \ddangerexercise What token list results from `|\uppercase{a\lowercase{bC}}|'\thinspace? \answer First |\uppercase| produces `|A\lowercase{BC}|'; then you get `|Abc|'. \ddangerexercise \TeX\ has an internal integer parameter called ^|\year| that is set equal to the current year number at the beginning of every job. Explain how to use |\year|, together with |\romannumeral| and |\uppercase|, to print a copyright notice like \year=1986 `\copyright\ \uppercase\expandafter{\romannumeral\year}' for all jobs run in \number\year. \answer `\thinspace|\copyright\ \uppercase\expandafter{\romannumeral\year}|% \thinspace'. \ (This is admittedly tricky; the `^|\expandafter|' expands the token after the `|{|', not the token after the group.) \ddangerexercise Define a control sequence |\appendroman| with three parameters such that |\appendroman#1#2#3| defines control sequence |#1| to expand to a control sequence whose name is the name of control sequence |#2| followed by the value of the positive integer |#3| expressed in roman numerals. For example, suppose |\count20| equals 30; then `|\appendroman\a\TeX{\count20}|' should have the same effect as `|\def\a{\TeXxxx}|'.^^{tricky macros} \answer (We assume that parameter |#2| is not simply an active character, and that ^|\escapechar| is between 0 and~255.) \begintt \def\gobble#1{} % remove one token \def\appendroman#1#2#3{\expandafter\def\expandafter#1\expandafter {\csname\expandafter\gobble\string#2\romannumeral#3\endcsname}} \endtt \endchapter Some bookes are to bee tasted, others to bee swallowed, and some few to bee chewed and disgested. \author FRANCIS ^{BACON}, {\sl Essayes\/} (1597) % p2 of orig edition \bigskip `Tis the good reader that makes the good book. \author RALPH WALDO ^{EMERSON}, {\sl Society \& Solitude\/} (1870) % Success \eject \beginchapter Chapter 8. The Characters\\You Type A lot of different keyboards are used with \TeX, but few keyboards can produce 256 different symbols. Furthermore, as we have seen, some of the characters that you {\sl can\/} type on your ^{keyboard} are reserved for ^^{terminal keyboard} special purposes like escaping and grouping. Yet when we studied fonts it was pointed out that there are 256 characters per font. So how can you refer to the characters that aren't on your keyboard, or that have been pre-empted for formatting? One answer is to use control sequences. For example, the plain format of Appendix B\null, which defines |%| to be a special kind of symbol so that you can use it for comments, defines the control sequence |\%| to mean a ^{percent sign}. To get access to any character whatsoever, you can type \begindisplay |\char|\ \enddisplay where \ is any number from 0 to 255 (optionally followed by a space); you will get the corresponding character from the current font. That's how Appendix~B handles |\%|; it defines `|\%|' to be an abbreviation for `|\char37|', since 37 is the character code for a percent sign. The codes that \TeX\ uses internally to represent characters are based on ``^{ASCII},'' the American Standard Code for Information Interchange. ^^{internal character codes} ^^{character codes} Appendix~C gives full details of this code, which assigns numbers to certain control functions as well as to ordinary letters and punctuation marks. For example, ^\${}=32$ and ^\${}=13$. There are 94~standard visible symbols, and they have been assigned code numbers from 33 to~126, inclusive. It turns out that `|b|' is character number 98 in ASCII. So you can typeset the word |bubble| in a strange way by putting \begintt \char98 u\char98\char98 le \endtt into your manuscript, if the |b|-key on your keyboard is broken. \ (An optional space is ignored after constants like `|98|'. Of course you need the |\|, |c|, |h|, |a|, and~|r| keys to type `^|\char|', so let's hope that they are always working.) \danger \TeX\ always uses the internal character code of Appendix~C for the standard ASCII characters, regardless of what external coding scheme actually appears in the files being read. Thus, |b| is 98 inside of \TeX\ even when your computer normally deals with ^{EBCDIC} or some other non-ASCII scheme; the \TeX\ software has been set up to convert text files to internal code, and to convert back to the external code when writing text files. Device-independent (^|dvi|) output files use \TeX's internal code. In this way, \TeX\ is able to give identical results on all computers. \danger Character code tables like those in Appendix~C often give the code numbers in {\sl ^{octal notation}}, i.e., the radix-8 number system, in which the digits are {\it0},~{\it1}, {\it2}, {\it3}, {\it4}, {\it5}, {\it6}, and~{\it7}.\footnote*{The author of this manual likes to use italic digits for octal numbers, and typewriter type for hexadecimal numbers, in order to provide a typographic clue to the underlying radix whenever possible.} Sometimes {\sl^{hexadecimal notation}\/} is also used, in which case the digits are |0|,~|1|, |2|, |3|, |4|, |5|, |6|, |7|, |8|, |9|, |A|, |B|, |C|, |D|, |E|, and~|F|. For example, the octal code for `|b|' is {\it142}, and its hexadecimal code is |62|. A ^\ in \TeX's language can begin with~a~|'|, in which case it is regarded as octal, or with a |"|, when it is regarded as hexadecimal. Thus, |\char'142| and |\char"62| are equivalent to |\char98|. The legitimate character codes in octal notation run from \oct0 to \oct{377}; in hexadecimal, they run from \hex0 to \hex{FF}. ^^{apostrophe}^^{doublequote} \danger But \TeX\ actually provides another kind of \ that makes it unnecessary for you to know ASCII at all! The token |`|$_{12}$ (^{left quote}), when followed by any character token or by any control sequence token whose name is a single character, stands for \TeX's internal code for the character in question. For example, |\char`b| and |\char`\b| are also equivalent to |\char98|. ^^{reverse apostrophe} If you look in Appendix~B to see how |\%| is defined, you'll notice that the definition is \begintt \def\%{\char`\%} \endtt instead of\/ |\char37| as claimed above. \dangerexercise What would be wrong with |\def\%{\char`%}|? \answer The |%| would be treated as a comment character, because its category code is~14; thus, no |%| token or |}| token would get through to the gullet of \TeX\ where numbers are treated. When a character is of category 0, 5, 9, 14, or~15, the extra |\| must be used; and the |\| doesn't hurt, so you can always use it to be safe. \ddanger The preface to this manual points out that the author tells little white lies from time to time. Well, if you actually check Appendix~B you'll find that \begintt \chardef\%=`\% \endtt is the true definition of\/ |\%|. Since format designers often want to associate a special character with a special control sequence name, \TeX\ provides the construction `^|\chardef|\|=|\' for numbers between 0 and 255, as an efficient alternative to `^|\def|\|{\char|\|}|'. Although you can use |\char| to access any character in the current font, you can't use it in the middle of a control sequence. For example, if you type \begintt \\char98 \endtt \TeX\ reads this as the control sequence |\\| followed by |c|, |h|, |a|, etc., not as the control sequence |\b|. You will hardly ever need to use |\char| when typing a manuscript, since the characters you want will probably be available as predefined control sequences; |\char| is primarily intended for the designers of book formats like those in the appendices. But some day you may require a ^{special symbol}, and you may have to hunt through a font catalog until you find it. Once you find it, you can use it by simply selecting the appropriate font and then specifying the character number with |\char|. For example, the ``^{dangerous bend}'' sign used in this manual appears as character number~127 of font ^|manfnt|, and that font is selected by the control sequence ^|\manual|. The macros in Appendix~E therefore display dangerous bends by saying `|{\manual\char127}|'. We have observed that the ASCII character set includes only 94 printable symbols; but \TeX\ works internally with 256 different character codes, from 0 to 255, each of which is assigned to one of the sixteen categories described in Chapter~7. If your keyboard has additional symbols, or if it doesn't have the standard~94, the people who installed your local \TeX\ system can tell you the correspondence between what you type and the character number that \TeX\ receives. Some people are fortunate enough to have keys marked `{\tentex\char'32}' and `{\tentex\char'34}' and `{\tentex\char'35}'; it is possible to install \TeX\ so that it will recognize these handy symbols and make the typing of mathematics more pleasant. But if you do not have such keys, you can get by with the control sequences ^|\ne|, ^|\le|, and ^|\ge|. ^^{not-equal}^^{less-or-equal}^^{greater-or-equal} \danger \TeX\ has a standard way to refer to the invisible characters of ASCII: Code~0 can be typed as the sequence of three characters |^^@|, code~1 can be typed |^^A|, and so on up to code~31, which is |^^_| (see Appendix~C\null). If the character following |^^| has an internal code between 64 and 127, \TeX\ subtracts 64 from the code; if the code is between 0 and 63, \TeX\ adds~64. Hence code 127 can be typed |^^?|, and the dangerous bend sign can be obtained by saying |{\manual^^?}|. However, you must change the category code of character 127 before using it, since this character ordinarily has category~15 (^{invalid}); say, e.g., |\catcode`\^^?=12|. ^^{double hat} ^^{hat hat} The |^^| notation is different from |\char|, because |^^| combinations are like single characters; for example, it would not be permissible to say |\catcode`\char127|, but |^^| symbols can even be used as letters within control words. \danger One of the overfull box messages in Chapter 6 illustrates the fact that \TeX\ sometimes uses the funny |^^| convention in its output: The umlaut character in that example appears as |^^?|, and the cedilla appears as~|^^X|, because `\thinspace\"{}\thinspace' and `\char'30' occur in positions \oct{177} and~\oct{30} of the ^|\tenrm| font. \danger There's also a special convention in which |^^| is followed by {\sl two\/} ``lowercase hexadecimal digits,'' |0|--|9| or |a|--|f|. With this convention, all 256 characters are obtainable in a uniform way, from |^^00| to |^^ff|. Character 127 is |^^7f|. \danger Most of the |^^| codes are unimportant except in unusual applications. But |^^M| is particularly noteworthy because it is code 13, the ASCII ^\ that \TeX\ normally places at the right end of every line of your input file. By changing the category of~|^^M| you can obtain useful special effects, as we shall see later. ^^{hat hat M} \danger The control code |^^I| is also of potential interest, since it's the ASCII ^\. Plain \TeX\ makes \ act like a blank space. \ddanger People who install \TeX\ systems for use with non-American alphabets can make \TeX\ conform to any desired standard. For example, suppose you have a ^{Norwegian keyboard} containing the letter {\tt\ae}, which ^^{Scandinavian letters} ^^{foreign languages} comes in as code~241 (say). Your local format package should define |\catcode`|{\tt\ae}|=11|; then you could have control sequences like |\s|{\tt\ae}|rtrykk|. Your \TeX\ input files could be made readable by American installations of \TeX\ that don't have your keyboard, by substituting |^^f1| for character~241. \ (For example, the stated control sequence would appear as |\s^^f1rtrykk| in the file; your American friends should also be provided with the format that you used, with its |\catcode`^^f1=11|.) \ Of course you should also arrange your fonts so that \TeX's character 241 will print as {\ae}; and you should change \TeX's hyphenation algorithm so that it will do correct Norwegian hyphenation. The main point is that such changes are not extremely difficult; nothing in the design of \TeX\ limits it to the American alphabet. Fine printing is obtained by fine tuning to the language or languages being used. ^^{keyboards, non-ASCII} \ddanger European languages can also be accommodated effectively with only a limited character set. For example, let's consider Norwegian again, but suppose that you want to use a keyboard without an {\tt\ae} character. You can arrange the ^{font metric file} so that \TeX\ will interpret |ae|, |o/|, |aa|, |AE|, |O/|, and |AA| as ligatures that produce \ae, \o, \aa, \AE, \O, and \AA, respectively; and you could put the characters \aa\ and \AA\ into positions 128 and~129 of the font. By setting |\catcode`/=11| you would be able to use the ligature |o/| in control sequences like `|\ho/yre|'. \TeX's hyphenation method is not confused by ligatures; so you could use this scheme to operate essentially as suggested before, but with two keystrokes occasionally replacing one. \ (Your typists would have to watch out for the occasional times when the adjacent characters |aa|, |ae|, and |o/| should not be treated as ligatures; also, `|\/|' would be a ^{control word}, not a ^{control symbol}.) \ddanger The rest of this chapter is devoted to \TeX's reading rules, which define the conversion from text to tokens. For example, the fact that \TeX\ ignores spaces after control words is a consequence of the rules below, which imply among other things that spaces after control words never become space tokens. The rules are intended to work the way you would expect them to, so you may not wish to bother reading them; but when you are communicating with a computer, it is nice to understand what the machine thinks it is doing, and here's your chance. \ddanger The input to \TeX\ is a sequence of ``^{lines}.'' Whenever \TeX\ is reading a line of text from a file, or a line of text that you entered directly on your terminal, the computer's reading apparatus is in one of three so-called ^{states}: \begindisplay \noalign{\vskip1pt} State $N$&Beginning a new line;\cr State $M$&Middle of a line;\cr State $S$&Skipping blanks.\cr \noalign{\vskip-3pt} \enddisplay At the beginning of every line it's in state $N$; but most of the time it's in state $M$, and after a control word or a space it's in state $S$. Incidentally, ``states'' are different from the ``^{modes}'' that we will be studying later; the current {\sl state\/} refers to \TeX's eyes and mouth as they take in characters of new text, but the current {\sl mode\/} refers to the condition of \TeX's gastro-intestinal tract. Most of the things that \TeX\ does when it converts characters to ^{tokens} are independent of the current state, but there are differences when spaces or end-of-line characters are detected (categories 10 and 5). \ddanger \TeX\ deletes any ^\ characters (number 32) that occur at the right end of an input line. Then it inserts a ^\ character (number~13) at the right end of the line, except that it places nothing additional at the end of a line that you inserted with `|I|' during ^{error recovery}. Note that \ is considered to be an actual character that is part of the line; you can obtain special effects by changing its catcode. \ddanger If \TeX\ sees an escape character (category 0) in any state, it scans the entire ^{control sequence} name as follows. (a)~If there are no more characters in the line, the name is empty (like |\csname\endcsname|). ^^{null control sequence} ^^{csname endcsname} Otherwise (b)~if the next character is not of category~11 (letter), the name consists of that single symbol. Otherwise (c)~the name consists of all letters beginning with the current one and ending just before the first nonletter, or at the end of the line. This name becomes a control sequence token. \TeX\ goes into state~$S$ in case~(c), or in case~(b) with respect to a character of category~10 (space); otherwise \TeX\ goes into state~$M$. \ddanger If \TeX\ sees a superscript character (category 7) in any state, and if that character is followed by another identical character, and if those two equal characters are followed by a character of code $c<128$, then they are deleted and 64 is added~to or subtracted from the code~$c$. \ (Thus, |^^A| is replaced by a single character whose code is~1, etc., as explained earlier.) \ However, if the two superscript characters are immediately followed by two of the lowercase hexadecimal digits |0123456789abcdef|, the four-character sequence is replaced by a single character having the specified hexadecimal code. The replacement is carried out also if such a trio or quartet of characters is encountered during steps (b) or~(c) of the control-sequence-name scanning procedure described above. After the replacement is made, \TeX\ begins again as if the new character had been present all the time. If a superscript character is not the first of such a trio or quartet, it is handled by the following rule. \ddanger If \TeX\ sees a character of categories 1, 2, 3, 4, 6, 8, 11, 12, or~13, or a character of category~7 that is not the first of a special sequence as just described, it converts the character to a token by attaching the category code, and goes into state~$M$. This is the normal case; almost every nonblank character is handled by this rule. \ddanger If \TeX\ sees an end-of-line character (category 5), it throws away any other information that might remain on the current line. Then if \TeX\ is in state~$N$ (new line), the end-of-line character is converted to the control sequence token `\cstok{par}' ^^|\par| (end of paragraph); if \TeX\ is in state~$M$ (mid-line), the end-of-line character is converted to a token for character~32 (`\]') of category~10 (^{space}); and if \TeX\ is in state~$S$ (skipping blanks), the end-of-line character is simply dropped. \ddanger If \TeX\ sees a character to be ignored (category~9), it simply bypasses that character as if it weren't there, and remains in the same state. \ddanger If \TeX\ sees a character of category~10 (space), the action depends on the current state. If \TeX\ is in state $N$ or $S$, the character is simply passed by, and \TeX\ remains in the same state. Otherwise \TeX\ is in state $M$; the character is converted to a token of category~10 whose character code is~32, and \TeX\ enters state~$S$. The character code in a space token is always~32. \ddanger If \TeX\ sees a comment character (category~14), it throws away that character and any other information that might remain on the current line. \ddanger Finally, if \TeX\ sees an invalid character (category~15), it bypasses that character, prints an error message, and remains in the same state. \ddanger If \TeX\ has nothing more to read on the current line, it goes to the next line and enters state $N$. However, if\/ ^|\endinput| has been specified for a file being ^|\input|, or if an |\input| file has ended, \TeX\ returns to whatever it was reading when the |\input| command was originally given. \ (Further details of\/ |\input| and |\endinput| are discussed in Chapter~20.) \ddangerexercise Test your understanding of \TeX's reading rules by answering the following quickie questions: (a)~What is the difference between categories 5 and~14? (b)~What is the difference between categories 3 and~4? (c)~What is the difference between categories 11 and~12? (d)~Are spaces ignored after active characters? (e)~When a line ends with a comment character like |%|, are spaces ignored at the beginning of the next line? (f)~Can an ignored character appear in the midst of a control sequence name? \answer (a)~Both characters terminate the current line; but a character of category~5 might be converted into \]$_{10}$ or a \cstok{par} token, while a character of category~14 never produces a token. (b)~They produce character tokens stamped with different category numbers. For example, |$|$_3$ is not the same token as |$|$_4$, so \TeX's digestive processes will treat them differently. (c)~Same as~(b), plus the fact that control sequence names treat letters differently. (d)~No. (e)~Yes; characters of category~10 are ignored at the beginning of every line, since every line starts in state~$N$. (f)~No. \ddangerexercise Look again at the error message that appears on page \vshippage. When \TeX\ reported that |\vship| was an undefined control sequence, it printed two lines of context, showing that it was in the midst of reading line~2 of the |story| file. At the time of that error message, what state was \TeX\ in? What character was it about to read next? \answer \TeX\ had just read the control sequence |\vship|, so it was in state~$S$, and it was just ready to read the space before `|1in|'. Afterwards it ignored that space, since it was in state~$S$; but if you had typed |I\obeyspaces| in response to that error message, you would have seen the space. Incidentally, when \TeX\ prints the ^{context of an error message}, the bottom pair of lines comes from a text file, but the other pairs of lines are portions of token lists that \TeX\ is reading (unless they begin with `|<*>|', when they represent text inserted during ^{error recovery}). \ddangerexercise Given the category codes of plain \TeX\ format, what tokens are produced from the input line `| $x^2$~ \TeX ^^62^^6|'\thinspace? \answer |$|$_{3}$ |x|$_{11}$ |^|$_7$ |2|$_{12}$ |$|$_{3}$ |~|$_{13}$ \]$_{10}$ \cstok{TeX} |b|$_{11}$ |v|$_{11}$ \]$_{10}$. The final space comes from the \ placed at the end of the line. Code |^^6| yields |v| only when not followed by |0|--|9| or |a|--|f|. The initial space is ignored, because state~$N$ governs the beginning of the line. \ddangerexercise Consider an input file that contains exactly three lines; the first line says `|Hi!|', while the other two lines are completely blank. What tokens are produced when \TeX\ reads this file, using the category codes of plain \TeX\ format? \answer |H|$_{11}$ |i|$_{11}$ |!|$_{12}$ \]$_{10}$ \cstok{par} \cstok{par}. The `\]' comes from the \ at the end of the first line; the second and third lines each contribute a \cstok{par}. \ddangerexercise Assume that the category codes of plain \TeX\ are in force, except that the characters |^^A|, |^^B|, |^^C|, |^^M| belong respectively to categories 0, 7, 10, and 11. What tokens are produced from the (rather ridiculous) input line `|^^B^^BM^^A^^B^^C^^M^^@\M|\]'? (Remember that this line is followed by \, which is |^^M|; and recall that |^^@| denotes the ^\ character, which has category~9 when |INITEX| begins.) \answer The two |^^B|'s are not recognized as consecutive superscript characters, since the first |^^B| is converted to code~2 which doesn't equal the following character |^|. Hence the result is seven tokens: |^^B|$_7$ |^^B|$_7$ |M|$_{11}$ \cstok{\^{}\^{}B} \]$_{10}$ |^^M|$_{11}$ \cstok{M\^{}\^{}M}. The last of these is a control word whose name has two letters. The \ after |\M| is deleted before \TeX\ inserts the \ token. \ddanger The special character inserted at the end of each line needn't be ^\; \TeX\ actually inserts the current value of an integer parameter called ^|\endlinechar|, which normally equals~13 but it can be changed like any other parameter. If the value of\/ |\endlinechar| is negative or greater than~255, no character is appended, and the effect is as if every line ends with~|%| (i.e., with a comment character). \ddanger Since it is possible to change the category codes, \TeX\ might actually use several different categories for the same character on a single line. For example, Appendices D and~E contain several ways to coerce \TeX\ to process text ``^{verbatim},'' so that the author could prepare this manual without great difficulty. \ (Try to imagine typesetting a \TeX\ manual; backslashes and other special characters need to switch back and forth between their normal categories and category~12!) \ Some care is needed to get the timing right, but you can make \TeX\ behave in a variety of different ways by judiciously changing the categories. On the other hand, it is best not to play with the category codes very often, because you must remember that characters never change their categories once they have become tokens. For example, when the arguments to a macro are first scanned, they are placed into a token list, so their categories are fixed once and for all at that time. The author has intentionally kept the category codes numeric instead of mnemonic, in order to discourage people from making extensive use of\/ |\catcode| changes except in unusual circumstances. \ddangerexercise Appendix B defines ^|\lq| and ^|\rq| to be abbreviations for |`| and |'| (single left and right quotes, respectively). Explain why the definitions \begintt \chardef\lq=96 \chardef\rq=39 \endtt would not be as good. \answer Both alternatives work fine in text; in particular, they combine as in |\lq\lq| to form ligatures. But the definition in Appendix~B works also in connection with constants; e.g., |\char\lq\%| and |\char\rq140| are valid. \ (Incidentally, the construction |\let\lq=`| would not work with constants, since the quotes in a ^\ must come from character tokens of category~12; after |\let\lq=`| the control sequence token |\lq| will not expand into a character token, nor {\sl is\/} it a character token!) ^^|\let| ^^{implicit character} \endchapter for life's not a paragraph \quad % he left a blank line here, really And death i think is no parenthesis. \author e.~e.~^{cummings}, {\sl since feeling is first\/} (1926) \bigskip This coded character set is to facilitate the general interchange of information among information processing systems, communication systems, and associated equipment. $\ldots$ An 8-bit set was considered but the need for more than 128 codes in general applications was not yet evident. \author ASA SUBCOMMITTEE X3.2, {\sl American Standard\break % Code for Information Interchange\/^^{ASCII}} (1963) % in {\sl Communications of the ACM\/} \eject \beginchapter Chapter 9. \TeX's Roman Fonts When you're typing a manuscript for \TeX, you need to know what symbols are available. The plain \TeX\ format of Appendix~B is based on the Computer Modern fonts, which provide the characters needed to typeset a wide variety of documents. It's time now to discuss what a person can do with plain \TeX\ when typing straight text. We've already touched on some of the slightly subtle things---for example, dashes and quotation marks were considered in Chapter~2, and certain kinds of accents appeared in the examples of Chapters 3 and~6. The purpose of this chapter is to give a more systematic summary of the possibilities, by putting all the facts together. Let's begin with the rules for the normal roman font (|\rm| or |\tenrm|); plain \TeX\ will use this font for everything unless you specify otherwise. Most of the ordinary symbols that you need are readily available and you can type them in the ordinary way: There's nothing special about \begindisplay \openup1pt the ^{letters} |A| to |Z| and |a| to |z|\cr the ^{digits} |0| to |9|\cr common ^{punctuation} marks |: ; ! ? ( ) [ ] ` ' - * / . , @|\cr \enddisplay except that \TeX\ recognizes certain combinations as ^{ligatures}: $$\openup1pt\halign{\indent#\hfil\cr |ff| yields ff\thinspace;$\!$\quad |fi| yields fi\thinspace;$\!$\quad |fl| yields fl\thinspace;$\!$\quad |ffi| yields ffi\thinspace;$\!$\quad |ffl| yields ffl\thinspace;\cr |``| yields``\thinspace;\qquad |''| yields ''\thinspace;\qquad |!||`| yields !`\thinspace;\qquad |?||`| yields ?`\thinspace;\cr |--| yields --\thinspace;\qquad |---| yields ---\thinspace.\cr}$$ ^^{Spanish ligatures} You can also type |+| and |=|, to get the corresponding symbols + and~=; but it's much better to use such characters only in math mode, i.e., enclosed between two |$| signs, since that tells \TeX\ to insert the proper spacing for mathematics. Math mode is explained later; for now, it's just a good idea to remember that formulas and text should be segregated. A non-mathematical hyphen and a non-mathematical slash should be specified by typing `|-|' and `|/|' outside of mathematics mode, but subtraction and division should be specified by typing `|-|' and `|/|' between |$|~signs. ^^{Colon} ^^{Semicolon} ^^{Exclamation point}^^{Shriek, see exclamation point} ^^{Question mark} ^^{Parentheses} ^^{Brackets} ^^{Apostrophe} ^^{Reverse apostrophe} ^^{Hamza, see apostrophe} ^^{Ain, see reverse apostrophe} ^^{Hyphen} ^^{Dash} ^^{Asterisk} ^^{At sign} ^^{Virgule, see slash} ^^{Solidus, see slash} ^^{Shilling sign, see slash} ^^{Slash} ^^{Period} ^^{Full stop, see period} ^^{Comma} ^^{Plus sign} ^^{Equals sign} The previous paragraph covers 80 of the 94 visible characters of standard ASCII; so your keyboard probably contains at least 14 more symbols, and you should learn to watch out for the remaining ones, since they are special. Four of these are pre\"empted by plain \TeX; if your manuscript requires the symbols \begintt $ # % & \endtt ^^{dollar sign} ^^{sharp sign, see hash mark} ^^{number sign, see hash mark} ^^{hash mark} ^^{percent sign} ^^{ampersand} you should remember to type them as \begintt \$ \# \% \& \endtt respectively. Plain \TeX\ also reserves the six symbols \begintt \ { } ^ _ ~ \endtt ^^{backslash} ^^{braces} ^^{curly braces, see braces} ^^{hat, see circumflex} ^^{circumflex} ^^{underline} ^^{tilde} but you probably don't mind losing these, since they don't appear in normal copy. Braces and backslashes are available via control sequences in math mode. \goodbreak There are four remaining special characters in the standard ASCII set: \begintt " || < > \endtt Again, you don't really want them when you're typesetting text. \ (Double-quote marks should be replaced either by |``| or by |''|; vertical lines and relation signs are needed only in math mode.) ^^{double-quote mark} ^^{vertical line, see norm} ^^{norm symbol} ^^{less than sign} ^^{greater than sign} Scholarly publications in English often refer to other languages, so plain \TeX\ makes it possible to typeset the most commonly used ^{accents}: $$\halign{\indent\hbox to 50pt{#\hfil}&\hbox to 35pt{#\hfil}&#\hfil\cr \it\negthinspace Type&\it to get\cr \noalign{\smallskip} |\`o|&\`o&(grave accent)\cr |\'o|&\'o&(acute accent)\cr |\^o|&\^o&(circumflex or ``hat'')\cr |\"o|&\"o&(umlaut or dieresis)\cr |\~o|&\~o&(tilde or ``squiggle'')\cr |\=o|&\=o&(macron or ``bar'')\cr |\.o|&\.o&(dot accent)\cr |\u o|&\u o&(breve accent)\cr |\v o|&\v o&(h\'a\v cek or ``check'')\cr |\H o|&\H o&(long Hungarian umlaut)\cr |\t oo|&\t oo&(tie-after accent)\cr}$$ ^^|\`| ^^{grave accent} ^^|\'| ^^{acute accent} ^^{esc hat} ^^{circumflex accent} ^^{hat accent} ^^|\"| ^^{umlaut accent} ^^{dieresis} ^^{esc tilde} ^^{tilde accent} ^^{squiggle accent} ^^|\=| ^^{macron accent} ^^{bar accent} ^^|\.| ^^{dot accent} ^^|\v| ^^{h\'a\v cek accent} ^^{check accent} ^^|\u| ^^{breve accent} ^^|\H| ^^{Hungarian umlaut} ^^|\t| ^^{tie-after accent} ^^{embellished letters, see accents} Within the font, such accents are designed to appear at the right height for the letter `o'; but you can use them over any letter, and \TeX\ will raise an accent that is supposed to be taller. Notice that spaces are needed in the last four cases, to separate the control sequences from the letters that follow. You could, however, type `|\H{o}|' in order to avoid putting a space in the midst of a word. \medbreak Plain \TeX\ also provides three accents that go underneath: $$\halign{\indent\hbox to 50pt{#\hfil}&\hbox to 35pt{#\hfil}&#\hfil\cr \it\negthinspace Type&\it to get\cr \noalign{\smallskip} |\c o|&\c o&(cedilla accent)\cr |\d o|&\d o&(dot-under accent)\cr |\b o|&\b o&(bar-under accent)\cr}$$ ^^|\c| ^^{cedilla accent} ^^|\d| ^^{dot-under accent} ^^{emphatics, see dot-under} ^^|\b| ^^{bar-under accent} And there are a few special letters: $$\halign{\indent\hbox to 50pt{#\hfil}&\hbox to 35pt{#\hfil}&#\hfil\cr \it\negthinspace Type&\it to get\cr \noalign{\smallskip} |\oe,\OE|&\oe,\thinspace\OE&(French ligature OE)\cr |\ae,\AE|&\ae,\thinspace\AE&(Latin ligature and Scandinavian letter AE)\cr |\aa,\AA|&\aa,\thinspace\AA&(Scandinavian A-with-circle)\cr |\o,\O|&\o,\thinspace\O&(Scandinavian O-with-slash)\cr |\l,\L|&\l,\thinspace\L&(Polish suppressed-L)\cr |\ss|&\ss&(German ``es-zet'' or sharp S)\cr}$$ ^^{Scandinavian letters} ^^{sharp S} ^^{es-zet} ^^{German} ^^{Polish} ^^{Norwegian} ^^{Danish} ^^{Swedish} ^^{suppressed-L} ^^{diphthongs, see \ae, \oe} The |\rm| font contains also the ^{dotless letters} `\i' and `\j', which you can obtain by typing `^|\i|' and `^|\j|'. These are needed because `i' and `j' should lose their dots when they gain an accent. For example, the right way to obtain `m\=\i n\u us' is to type \hbox{`|m\=\i n\u us|'} or `|m\={\i}n\u{u}s|'. This completes our summary of the |\rm| font. Exactly the same conventions apply to |\bf|, |\sl|, and |\it|, so you don't have to do things differently when you're using a different typeface. For example, |\bf\"o| yields {\bf\"o} and |\it\&| yields {\it\&}. Isn't that nice? \danger However, |\tt| is slightly different. You will be glad to know that |ff|, |fi|, and so on are not treated as ligatures when you're using ^{typewriter type}; nor do you get ligatures from dashes and quote marks. That's fine, because ordinary dashes and ordinary double-quotes are appropriate when you're trying to imitate a typewriter. Most of the accents are available too. But |\H|, |\.|, |\l|, and |\L| cannot be used---the typewriter font contains other symbols in their place. Indeed, you are suddenly allowed to type |"|, \|, |<|, and |>|; ^^{doublequote} ^^{vertical line} ^^{less than sign} ^^{greater than sign} see Appendix~F\null. All of the letters, spaces, and other symbols in |\tt| have the same width. \exercise What's the non-naive way to type `na\"\i ve'\thinspace? \answer |na\"\i ve| or |na{\"\i}ve| or |na\"{\i}ve|. \exercise List some English words that contain accented letters. \answer Belov\`ed prot\'eg\'e; r\^ole co\"ordinator; souffl\'es, cr\^epes, p\^at\'es, etc. \exercise How would you type `\AE sop's \OE uvres en fran\c cais'\thinspace? \answer |\AE sop's \OE uvres en fran\c cais|. \exercise Explain what to type in order to get this sentence: {\sl Commentarii Academi\ae\ scientiarum imperialis petropolitan\ae\/} became {\sl Akademi\t\i a Nauk SSSR, Doklady}. \answer |{\sl Commentarii Academi\ae\ scientiarum imperialis|\hfil\break |petropolitan\ae\/} became {\sl Akademi\t\i a Nauk SSSR, Doklady}.| \exercise And how would you specify the names Ernesto ^{Ces\`aro}, P\'al ^{Erd\H os}, \O ystein ^{Ore}, Stanis\l aw \'Swierczkowski, ^^{Swiercz...} Serge\u\i\ \t Iur'ev, ^^{Iur'ev} Mu\d hammad ibn M\^us\^a ^{al-Khw\^arizm\^\i}? \answer |Ernesto Ces\`aro, P\'al Erd\H os, \O ystein Ore, Stanis\l aw \'Swier%|\break|czkowski, Serge\u\i\ \t Iur'ev, Mu\d hammad ibn M\^us\^a al-Khw\^arizm\^\i.| \dangerexercise Devise a way to typeset {\tt P\'al Erd{\bf\H{\tt o}}s} in typewriter type. \answer The proper umlaut is |\H|, which isn't available in |\tt|, so it's necessary to borrow the accent from another font. For example, \hbox{|{\tt P\'al Erd{\bf\H{\tt o}}s}|} uses a bold accent, which is suitably dark. The following symbols come out looking exactly the same whether you are using |\rm|, |\sl|, |\bf|, |\it|, or |\tt|: $$\halign{\indent#\hfil\ &\hfil#\hfil&#\hfil\cr \it\negthinspace Type&\it to get\cr \noalign{\smallskip} |\dag|&\dag&(dagger or obelisk)\cr |\ddag|&\ddag&(double dagger or diesis)\cr |\S|&\S&(section number sign)\cr |\P|&\P&(paragraph sign or pilcrow)\cr}$$ ^^{dagger} ^^{double dagger} ^^{obelisk} ^^{obelus, see obelisk} ^^{diesis} ^^{section number sign} ^^{paragraph sign} ^^{pilcrow, see paragraph sign} (They appear in just one style because plain \TeX\ gets them from the math symbols font. Lots of other symbols are needed for mathematics; we shall study them later. See Appendix~B for a few more non-math symbols.) \exercise In plain \TeX's italic font, the `\$' sign comes out as `{\it\$}\thinspace'. ^^{dollar sign} ^^{British pound sign} ^^{pound sterling} ^^{sterling} This gives you a way to refer to pounds sterling, but you might want an italic dollar sign. Can you think of a way to typeset a reference to the book {\it Europe on {\sl\$}15.00 a day}\thinspace? \answer |{\it Europe on {\sl\$}15.00 a day\/}| \ddanger Appendix B shows that plain \TeX\ handles most of the accents by using \TeX's ^|\accent| primitive. For example, |\'#1| is equivalent to |{\accent19 #1}|, where |#1| is the argument being accented. The general rule is that |\accent|\ puts an accent over the next character; the \ tells where that accent appears in the current font. The accent is assumed to be properly positioned for a character whose height equals the ^{x-height} of the current font; taller or shorter characters cause the accent to be raised or lowered, taking due account of the slantedness of the fonts of accenter and accentee. The width of the final construction is the width of the character being accented, regardless of the width of the accent. Mode-independent commands like font changes may appear between the accent number and the character to be accented, but grouping operations must not intervene. If it turns out that no suitable character is present, the accent will appear by itself as if you had said |\char|\ instead of\/ |\accent|\. For example, |\'{}| produces \'{}. \ddangerexercise Why do you think plain \TeX\ defines |\'#1| to be `|{\accent19 #1}|' instead of simply letting |\'| be an abbreviation for `|\accent19 |'\thinspace? \ (Why the extra braces, and why the argument |#1|?) \answer The extra braces keep font changes local. An argument makes the use of\/ |\'| more consistent with the use of other accents like |\d|, which are manufactured from other characters without using the |\accent| primitive. \ddanger It's important to remember that these conventions we have discussed for accents and special letters are not built into \TeX\ itself; they belong only to the plain \TeX\ format, which uses the Computer Modern fonts. Quite different conventions will be appropriate when other fonts are involved; format designers should provide rules for how to obtain accents and special characters in their particular systems. Plain \TeX\ works well enough when accents are infrequent, but the conventions of this chapter are by no means recommended for large-scale applications of \TeX\ to other languages. For example, a well-designed \TeX\ font for ^{French} might well treat accents as ligatures, so that one could |e'crire de cette manie`re nai"ve en franc/ais| without backslashes. (See the remarks about Norwegian in Chapter~8.) ^^{foreign languages} \endchapter Let's doo't after the high Roman fashion. \author WILLIAM ^{SHAKESPEARE}, {\sl The Tragedie of % Anthony and Cleopatra\/} (1606) % Act IV, Scene 13, line 87 \bigskip English is a straightforward, frank, honest, open-hearted, no-nonsense language, which has little truck with such devilish devious devices as accents; indeed U.S. editors and printers are often thrown into a dither when a foreign word insinuates itself into the language. However there is one word on which Americans seem to have closed ranks, printing it confidently, courageously, and almost invariably complete with accent---the cheese presented to us as M\"unster. \smallskip Unfortunately, ^{Munster} doesn't take an accent. \author WAVERLEY ^{ROOT}, in the {\sl International Herald Tribune\/} (1982) % Tuesday 18 May 82 page 8 \eject \beginchapter Chapter 10. Dimensions Sometimes you want to tell \TeX\ how big to make a space, or how wide to make a line. For example, the short story of Chapter~6 used the instruction `|\vskip .5cm|' to skip vertically by half a centimeter, and we also said `|\hsize=4in|' to specify a horizontal size of 4~inches. It's time now to consider the various ways such ^{dimensions} can be communicated to \TeX. ``^{Points}'' and ``^{picas}'' are the traditional units of measure for printers and compositors in English-speaking countries, so \TeX\ understands points and picas. \TeX\ also understands inches and metric units, as well as the continental European versions of points and picas. Each unit of measure is given a two-letter abbreviation, as follows: ^^{units of measure, table} $$\halign{\indent\tt#&\quad#\hfil\cr pt&point (baselines in this manual are $12\pt$ apart)\cr pc&pica ($\rm1\,pc=12\,pt$)\cr in&inch ($\rm1\,in=72.27\,pt$)\cr bp&big point ($\rm72\,bp=1\,in$)\cr cm¢imeter ($\rm2.54\,cm=1\,in$)\cr mm&millimeter ($\rm10\,mm=1\,cm$)\cr dd&didot point ($\rm1157\,dd=1238\,pt$)\cr cc&cicero ($\rm1\,cc=12\,dd$)\cr sp&scaled point ($\rm65536\,sp=1\,pt$)\cr}$$ ^^|pt|^^{point} ^^|pc|^^{pica} ^^|in|^^{inch} ^^|bp|^^{big point} ^^|cm|^^{centimeter} ^^|mm|^^{millimeter} ^^|dd|^^{didot point}^^{Didot, F. A.} ^^|cc|^^{cicero} ^^|sp|^^{scaled point} The output of \TeX\ is firmly grounded in the metric system, using the conversion factors shown here as exact ratios. \exercise How many points are there in 254 centimeters? \answer Exactly $7227\pt$. When you want to express some physical dimension to \TeX, type it as \begindisplay \\\\cr \noalign{\hbox{or}} \\|.|\\\cr \enddisplay where an ^\ is either a `|+|' or a `|-|' or nothing at all, and where a ^\ consists of zero or more consecutive decimal digits. The `|.|'\ can also be a `|,|'\null. For example, here are six typical dimensions: $$\halign{\indent#\hfil&\hskip 6em#\hfil\cr |3 in|&|29 pc|\cr |-.013837in|&|+ 42,1 dd|\cr |0.mm|&|123456789sp|\cr}$$ A plus sign is redundant, but some people occasionally like extra redundancy once in a~while. Blank spaces are optional before the signs and the numbers and the units of measure, and you can also put an optional space after the dimension; but you should not put spaces within the digits of a number or between the letters of the unit of measure. \exercise Arrange those six ``typical dimensions'' into order, from smallest to largest. \answer $\rm-.013837\,in$, $\rm0.\,mm$, $\rm+42.1\,dd$, $\rm3\,in$, $\rm29\,pc$, $\rm123456789\,sp$. \ (The lines of text in this manual are 29~picas wide.) \dangerexercise Two of the following three dimensions are legitimate according to \TeX's rules. Which two are they? What do they mean? Why is the other one incorrect? \begintt '.77pt "Ccc -,sp \endtt \answer The first is not allowed, since octal notation cannot be used with a decimal point. The second is, however, legal, since a \ can be hexadecimal according to the rule mentioned in Chapter~8; it means $\rm12\,cc$, which is $\rm144\,dd\approx154.08124\,pt$. The third is also accepted, since a \ can be empty; it is a complicated way to say $\rm0\,sp$. \smallskip The following ``rulers'' have been typeset by \TeX\ so that you can get some idea of how different units compare to each other. If no distortion has been introduced during the camera work and printing processes that have taken place after \TeX\ did its work, these rulers are highly accurate. $$ \abovedisplayskip 15pt plus 4pt minus 4pt \belowdisplayskip 15pt plus 4pt minus 4pt \vbox{ \def\1{\vrule height 0pt depth 2pt} \def\2{\vrule height 0pt depth 4pt} \def\3{\vrule height 0pt depth 6pt} \def\4{\vrule height 0pt depth 8pt} \def\ruler#1#2#3{\leftline{$\vcenter{\hrule\hbox{\4#1}}\,\,\rm#2\,{#3}$}} \def\\#1{\hbox to .125in{\hfil#1}} \def\8{\\\1\\\2\\\1\\\3\\\1\\\2\\\1\\\4} \ruler{\8\8\8\8}4{in} \vskip 18pt \def\\#1{\hbox to 10pt{\hfil#1}} \def\8{\\\1\\\1\\\1\\\1\\\2\\\1\\\1\\\1\\\1\\\4} \ruler{\8\8\8}{300}{pt} \vskip 18pt \def\\#1{\hbox to 10dd{\hfil#1}} \def\8{\\\1\\\1\\\1\\\1\\\2\\\1\\\1\\\1\\\1\\\4} \ruler{\8\8\8}{300}{dd} \vskip 18pt \def\\#1{\hbox to 5mm{\hfil#1}} \def\8{\\\2\\\4} \ruler{\8\8\8\8\8\8\8\8\8\8}{10}{cm} \vskip 6pt}$$ \dangerexercise (To be worked after you know about boxes and glue and have read Chapter~21.) \ Explain how to typeset such a $\rm10\,cm$ ^{ruler}, using \TeX. \answer {\obeylines|\def\tick#1{\vrule height 0pt depth #1pt}| |\def\\{\hbox to 1cm{\hfil\tick4\hfil\tick8}}| |\vbox{\hrule\hbox{\tick8\\\\\\\\\\\\\\\\\\\\}}| \noindent(You might also try putting ticks at every millimeter, in order % to see how good your system is; % some output devices can't handle 101~rules all at once.)} \danger \TeX\ represents all dimensions internally as an integer multiple of the tiny units called sp. Since the wavelength of visible light is approximately $\rm100\,sp$, % in fact: violet=75sp, red=135sp! rounding errors of a few sp make no difference to the eye. However, \TeX\ does all of its arithmetic very carefully so that identical results will be obtained on different computers. Different implementations of \TeX\ will produce the same line breaks and the same page breaks when presented with the same document, because the integer arithmetic will be the same. ^^{machine-independence} ^^{rounding} \danger The units have been defined here so that precise conversion to~sp is~efficient on a wide variety of machines. In order to achieve this, \TeX's ``pt'' has been made slightly larger than the official printer's point, which was defined to equal exactly $\rm.013837\,in$ by the American Typefounders Association in~1886 [cf.~National Bureau of Standards Circular~570 (1956)]. In fact, one classical point is exactly $.99999999\pt$, so the ``error'' is essentially one part in $10^8$. This is more than two orders of magnitude less than the amount by which the inch itself changed during 1959, when it shrank to $\rm2.54\,cm$ from its former value of $\rm(1/0.3937)\,cm$; so there is no point in worrying about the difference. The new definition $\rm72.27\,pt=1\,in$ is not only better for calculation, it~is also easier to remember. \danger \TeX\ will not deal with dimensions whose absolute value is $\rm2^{30}\,sp$ or more. In other words, the ^{maximum legal dimension} is slightly less than $16384\pt$. This is a distance of about 18.892 feet (5.7583 meters), so it won't cramp your style. In a language manual like this it is convenient to use ``^{angle brackets}'' in abbreviations for various constructions like \ and \ and \. Henceforth we shall use the term ^\ to stand for a legitimate \TeX\ dimension. For example, \begindisplay |\hsize=|\ \enddisplay will be the general way to define the column width that \TeX\ is supposed to use. The idea is that \ can be replaced by any quantity like `|4in|' that satisfies \TeX's grammatical rules for dimensions; abbreviations in angle brackets make it easy to state such laws of grammar. When a dimension is zero, you have to specify a unit of measure even though the unit is irrelevant. Don't just say `|0|'\thinspace; say `|0pt|' or `|0in|' or something. \smallbreak The 10-point size of type that you are now reading is normal in textbooks, but you probably will often find yourself wanting a larger font. Plain \TeX\ makes it easy to do this by providing {\magnifiedfiverm ^{magnif{}ied output}.} If you say \begintt \magnification=1200 \endtt at the beginning of your manuscript, everything will be enlarged by 20\%; i.e., it will come out at 1.2 times the normal size. Similarly, `|\magnification=2000|' doubles everything; this actually quadruples the area of each letter, since heights and widths are both doubled. To magnify a document by the factor $f$, you say ^|\magnification||=|\, where the \ is 1000~times~$f$. This instruction must be given before the first page of output has been completed. You cannot apply two different magnifications to the same document. Magnification has obvious advantages: You'll have less ^{eyestrain} when you're ^{proofreading}; you can easily make ^{transparencies} ^^{slides} for lectures; and you can photo-reduce magnified output, in order to minimize the deficiencies of a ^{low-resolution printer}. Conversely, you might even want `|\magnification=500|' in order to create a ^{pocket-size} version of some book. ^^{squint print} But there's a slight catch: You can't use magnification unless your printing device happens to have the fonts that you need at the magnification you desire. In other words, you need to find out what sizes are available before you can magnify. Most installations of \TeX\ make it possible to print all the fonts of plain \TeX\ if you magnify by ^|\magstep||0|,~|1|, |2|,~|3|, and perhaps~|4| or even~|5| (see Chapter~4); but the use of large fonts can be expensive because a lot of system memory space is often required to store the shapes. \exercise Try printing the short story of Chapter 6 at 1.2, 1.44, and 1.728 times the normal size. What should you type to get \TeX\ to do this? \answer For example, say `|\magnification=\magstep1 \input story \end|' to get magnification 1200; |\magstep2| and |\magstep3| are 1440 and 1728. Three separate runs are needed, since there can be at most one magnification per job. The output may look funny if the fonts don't exist at the stated magnifications. \danger When you say |\magnification=2000|, an operation like `|\vskip.5cm|' will actually skip $\rm1.0\,cm$ of space in the final document. If you want to specify a dimension in terms of the final size, \TeX\ allows you to say `^|true|' just before |pt|, |pc|, |in|, |bp|, |cm|, |mm|, |dd|, |cc|, and |sp|. This unmagnifies the units, so that the subsequent magnification will cancel out. For example, `|\vskip.5truecm|' is equivalent to `|\vskip.25cm|' if you have previously said `|\magnification=2000|'. Plain \TeX\ uses this feature in the |\magnification| command itself: Appendix~B includes the instruction \begintt \hsize = 6.5 true in \endtt just after a new magnification has taken effect. This adjusts the line width so that the material on each page will be $6{1\over2}$ inches wide when it is finally printed, regardless of the magnification factor. There will be an inch of margin at both left and right, assuming that the paper is $8{1\over2}$ inches wide. \danger If you use no `|true|' dimensions, \TeX's internal computations are not affected by the presence or absence of magnification; line breaks and page breaks will be the same, and the ^|dvi| file will change in only two places. \TeX\ simply tells the printing routine that you want a certain magnification, and the printing routine will do the actual enlargement when it reads the |dvi| file. \dangerexercise Chapter 4 mentions that fonts of different magnifications can be used in the same job, by loading them `^|at|' different sizes. Explain what fonts will be used when you give the commands ^^{magnified fonts} ^^|scaled| \begintt \magnification=\magstep1 \font\first=cmr10 scaled\magstep1 \font\second=cmr10 at 12truept \endtt \answer Magnification is by a factor of 1.2. Since font |\first| is |cmr10| at $12\pt$, it will be |cmr10| at $14.4\pt$ after magnification; font |\second| will be |cmr10| at $12\pt$. \ (\TeX\ changes `|12truept|' into `|10pt|', and the final output magnifies it back to $12\pt$.) \ddanger Magnification is actually governed by \TeX's ^|\mag| primitive, which is an integer parameter that should be positive and at~most~32768. The value of\/ |\mag| is examined in three cases: (1)~just before the first page is shipped to the |dvi| file; (2)~when computing a |true| dimension; (3)~when the |dvi| file is being closed. Alternatively, some implementations of \TeX\ produce non-|dvi| output; they examine |\mag| in case~(2) and also when shipping out each page. Since each document has only one magnification, the value of\/ |\mag| must not change after it has first been examined. \danger \TeX\ also recognizes two units of measure that are relative rather than absolute; i.e., they depend on the current context: \begindisplay ^|em| is the width of a ``^{quad}'' in the current font;\cr ^|ex| is the ``^{x-height}'' of the current font.\cr \enddisplay Each font defines its own em and ex values. In olden days, an ``em'' was the width of an `M', but this is no longer true; ems are simply arbitrary units that come with a font, and so are exes. The Computer Modern fonts have the property that an em-dash is one em wide, each of the ^{digits} 0 to~9 is half an em wide, and lowercase `x' is one ex high; but these are not hard-and-fast rules for all fonts. The |\rm| font (^|cmr10|) of plain \TeX\ has $\rm1\,em=10\,pt$ and $\rm1\,ex\approx4.3\,pt$; the |\bf| font (^|cmbx10|) has $\rm1\,em=11.5\,pt$ and $\rm1\,ex\approx4.44\,pt$; and the |\tt| font (^|cmtt10|) has $\rm1\,em=10.5\,pt$ and $\rm1\,ex\approx4.3\,pt$. All of these are ``10-point'' fonts, yet they have different em and ex values. It~is generally best to use |em| for horizontal measurements and |ex| for vertical measurements that depend on the current font. \danger A \ can also refer to \TeX's internal registers or parameters. We shall discuss registers later, and a complete definition of everything that a ^\ can be will be given in Chapter~24. For now it will suffice to give some hints about what is to come: `|\hsize|' stands for the current horizontal line size, and `|.5\hsize|' is half that amount; `|2\wd3|' denotes twice the width of register~|\box3|; `|-\dimen100|' is the negative of register~|\dimen100|. \ddanger Notice that the unit names in dimensions are not preceded by backslashes. The same is true of other so-called ^{keywords} of the \TeX\ language. Keywords can be given in uppercase letters or in a mixture of upper and lower case; e.g., `|Pt|' is equivalent to `|pt|'. The category codes of these letters are irrelevant; you may, for example, be using a |p| of category~12 (other) that was generated by expanding `|\the\hsize|' as explained in Chapter~20. \TeX\ gives a special interpretation to keywords only when they appear in certain very restricted contexts. For example, `|pt|' is a keyword only when it appears after a number in a \; `|at|' is a keyword only when it appears after the external name of a font in a |\font| declaration. Here is a complete list of \TeX's keywords, in case you are wondering about the full set: |at|, |bp|, |by|, |cc|, |cm|, |dd|, |depth|, |em|, |ex|, |fil|, |height|, |in|, |l|, |minus|, |mm|, |mu|, |pc|, |plus|, |pt|, |scaled|, |sp|, |spread|, |to|, |true|, |width|. ^^{reserved words} \ (See Appendix~I for references to the contexts in which each of these is recognized as a keyword.) \endchapter The methods that have hitherto been taken to discover the measure of the Roman foot, will, upon examination, be found so unsatisfactory, that it is no wonder the learned are not yet agreed on that point. $\ldots$ 9 London inches are equal to 8,447 Paris inches. \author MATTHEW ^{RAPER}, in {\sl Philosophical Transactions\/} (1760) % ``An Enquiry into the Measure of the {\sl Roman\/} Foot,'' % {\sl Philos.\ Trans.\ \bf51} (1760), 774--823. \bigskip \checkequals\sesame\pageno % Without the letter U, units would be nits. \author ^{SESAME STREET}{^^{Children's Television Workshop}} (1970) \eject \beginchapter Chapter 11. Boxes \TeX\ makes complicated pages by starting with simple individual characters and putting them together in larger units, and putting these together in still larger units, and so on. Conceptually, it's a big paste-up job. The \TeX nical terms used to describe such page construction are {\sl ^{boxes}\/} and {\sl ^{glue}}. Boxes in \TeX\ are two-dimensional things with a rectangular shape, having three associated measurements called {\sl^{height}}, {\sl^{width}}, and {\sl^{depth}}. Here is a picture of a typical box, showing its so-called ^{reference point} and ^{baseline}: {\eightpoint \setbox0=\hbox{$\uparrow$} \setbox1=\hbox to \wd0{$\hss\mid\hss$} % with luck, they'll line up \setbox2=\vbox{\copy0 \nointerlineskip \kern-.5pt \copy1 \nointerlineskip \kern-.5pt \copy1 \moveleft 1em\hbox{height} \copy1 \nointerlineskip \kern-.5pt \copy1 \nointerlineskip \kern-.5pt \hbox{$\downarrow$} \kern.2pt} \setbox3=\vbox{\kern.2pt\copy0 \moveleft 1em\hbox{depth} \hbox{$\downarrow$} \kern0pt} \setbox4=\vtop{\kern-3pt % this cancels the null text above the samplebox \hbox{\samplebox{\ht2}{\ht3}{6em}{}% \kern-6em \raise3pt\hbox to 6em{\hss Baseline\hss}} \kern3pt \arrows{6em}{width}} \medskip\indent \setbox0=\hbox{$\vcenter{}$}% \ht0 is the axis height \lower\ht0\hbox{Reference point$-$\kern-.2em$\rightarrow$\kern2pt}% \raise\ht2\box4 \kern1.5em \raise\ht2\vtop{\kern0pt\box2\nointerlineskip\box3}} \medskip\noindent From \TeX's viewpoint, a single character from a font is a box; it's one of the simplest kinds of boxes. The font designer has decided what the height, width, and depth of the character are, and what the symbol will look like when it is in the box; \TeX\ uses these dimensions to paste boxes together, and ultimately to determine the locations of the reference points for all characters on a page. In plain \TeX's |\rm| font (|cmr10|), for example, the letter `h' has a height of 6.9444 points, a width of 5.5555 points, and a depth of zero; the letter `g' has a height of 4.3055 points, a width of 5 points, and a depth of 1.9444 points. Only certain special characters like parentheses have height plus depth actually equal to 10 points, although ^|cmr10| is said to be a ``10-point'' font. You needn't bother to learn these measurements yourself, but it's good to be aware of the fact that \TeX\ deals with such information; then you can better understand what the computer does to your manuscript. The character shape need not fit inside the boundaries of its box. For example, some characters that are used to build up larger math symbols like matrix brackets intentionally protrude a little bit, so that they overlap properly with the rest of the symbol. Slanted letters frequently extend a little to the right of the box, as if the box were skewed right at the top and left at the bottom, keeping its baseline fixed. For example, compare the letter `g' in the |cmr10| and ^|cmsl10| fonts (|\rm| and |\sl|): \begindisplay \vbox to 40pt{\ifproofmode\hrule\vfill \hsize=2.5in \baselineskip 6pt \fiverm\noindent (A figure will be inserted here; too bad you can't see it now. It shows two g's, as claimed.) \vfill\hrule\fi} \enddisplay In both cases \TeX\ thinks that the box is 5 points wide, so both letters get exactly the same treatment. \TeX\ doesn't have any idea where the ink will go---only the output device knows this. But the slanted letters will be spaced properly in spite of \TeX's lack of knowledge, because the baselines will match up. Actually the font designer also tells \TeX\ one other thing, the so-called {\sl^{italic correction}\/}: A number is specified for each character, telling roughly how far that character extends to the right of its box boundary, plus a little to spare. For example, the italic correction for `g' in |cmr10| is $0.1389\pt$, while in |cmsl10| it is $0.8565\pt$. Chapter~4 points out that this correction is added to the normal width if you type `^|\/|' just after the character. You should remember to use |\/| when shifting from a slanted font to an unslanted one, especially in cases like \begintt the so-called {\sl italic correction\/}: \endtt since no space intervenes here to compensate for the loss of slant. \smallbreak \TeX\ also deals with another simple kind of box, which might be called a~``^{black box},'' namely, a rectangle like `\thinspace \vrule width 4pt height 6pt depth 1.5pt \thinspace' that is to be entirely filled with ink at printing time. You can specify any height, width, and depth you like for such boxes---but they had better not have too much area, or the printer might get upset. \ (Printers generally prefer white space to black space.) Usually these black boxes are made very skinny, so that they appear as horizontal lines or vertical lines. Printers traditionally call such lines ``^{horizontal rules}'' and ``^{vertical rules},'' so the terms \TeX\ uses to stand for black boxes are ^|\hrule| and ^|\vrule|. Even when the box is square, as in `\thinspace\bull\thinspace', you must call it either an~|\hrule| or a~|\vrule|. We shall discuss the use of ^{rule boxes} in greater detail later. \ (See Chapter~21.) \smallbreak Everything on a page that has been typeset by \TeX\ is made up of simple character boxes or rule boxes, pasted together in combination. \TeX\ pastes boxes together in two ways, either {\sl horizontally\/} or {\sl vertically}. When \TeX\ builds a ^{horizontal list} of boxes, it lines them up so that their reference points appear in the same horizontal row; therefore the baselines of adjacent characters will match up as they should. Similarly, when \TeX\ builds a ^{vertical list} of boxes, it lines them up so that their reference points appear in the same vertical column. % Here are some macros for making blank boxes \def\dolist{\afterassignment\dodolist\let\next= } \def\dodolist{\ifx\next\endlist \let\next\relax \else \\\let\next\dolist \fi \next} \def\endlist{\endlist} \def\\{\expandafter\if\space\next\ \else \setbox0=\hbox{\next}\maketypebox\fi} \def\demobox#1{\setbox0=\hbox{\dolist#1\endlist}% \copy0\kern-\wd0\makelightbox} Let's take a look at what \TeX\ does behind the scenes, by comparing the computer's methods with what you would do if you were setting metal type by hand. In the time-tested traditional method, you choose the letters that you~need out of a type case---the uppercase letters are in the ^{upper case}---and you put them into a ``^{composing stick}.'' When a line is complete, you adjust the spacing and transfer the result to the ``chase,'' where it joins the other rows of type. Eventually you lock the type up tightly by adjusting external wedges called ``quoins.'' This isn't much different from what \TeX\ does, except that different words are used; when \TeX\ locks up a line, it creates what is called an ``^{hbox}'' (^{horizontal box}), because the components of the line are pieced together horizontally. You can give an instruction like \begintt \hbox{A line of type.} \endtt in a \TeX\ manuscript; this tells the computer to take boxes for the appropriate letters in the current font and to lock them up in an hbox. As far as \TeX\ is concerned, the letter `A' is a box `\thinspace\setbox0\hbox{A}\maketypebox\thinspace' and the letter `p' is a box `\thinspace\setbox0\hbox{p}\maketypebox\thinspace'. So the given instruction causes \TeX\ to form the hbox \begindisplay \demobox{A line of type.} \enddisplay representing `A line of type.' The hboxes for individual lines of type are eventually joined together by putting them into a ``^{vbox}'' (^{vertical box}). For example, you can say \begintt \vbox{\hbox{Two lines}\hbox{of type.}} \endtt and \TeX\ will convert this into \begindisplay% \setbox0=\vbox{\hbox{\demobox{Two lines}}\hbox{\demobox{of type.}}} $\vcenter{\hbox{\makelightbox\kern-\wd0\box0}}$\qquad i.e.,\qquad$\vcenter{\vbox{\hbox{Two lines}\hbox{of type.}}}$ \enddisplay The principal difference between \TeX's method and the old way is that metal types are generally cast so that each character has the same height and depth; this makes it easy to line them up by hand. \TeX's types have variable height and depth, because the computer has no trouble lining characters up by their baselines, and because the extra information about height and depth helps in the positioning of accents and mathematical symbols. Another important difference between \TeX\ setting and hand setting is, of course, that \TeX\ will choose line divisions automatically; you don't have to insert ^|\hbox| and ^|\vbox| instructions unless you want to retain complete control over where each letter goes. On the other hand, if you do use |\hbox| and |\vbox|, you can make \TeX\ do almost everything that Ben ^{Franklin} could do in his printer's shop. You're only giving up the ability to make the letters come out charmingly crooked or badly inked; for such effects you need to make a new font. \ (And of course you lose the tactile and olfactory sensations, and the thrill of doing everything by yourself. \TeX\ will never completely replace the good~old~ways.) A page of text like the one you're reading is itself a box, in \TeX's view: It is a largish box made from a vertical list of smaller boxes representing the lines of text. Each line of text, in turn, is a box made from a horizontal list of boxes representing the individual characters. In more complicated situations, involving mathematical formulas and/or complex tables, you can have boxes within boxes within boxes $\ldots$ to any level. But even these complicated situations arise from horizontal or vertical lists of boxes pasted together in a simple way; all that you and \TeX\ have to worry about is one list of boxes at a time. In fact, when you're typing straight text, you don't have to think about boxes at all, since \TeX\ will automatically take responsibility for assembling the character boxes into words and the words into lines and the lines into pages. You need to be aware of the box concept only when you want to do something out of the ordinary, e.g., when you want to center a heading. \danger From the standpoint of \TeX's digestive processes, a manuscript comes in as a sequence of tokens, and the tokens are to be transformed into a sequence of boxes. Each token of input is essentially an instruction or a piece of an instruction; for example, the token `|A|$_{11}$' normally means, ``put a character box for the letter |A| at the end of the current hbox, using the current font''; the token `\cstok{vskip}' normally means, ``skip vertically in the current vbox by the \ specified in the following tokens.'' \danger The height, width, or depth of a box might be negative, in which case it is a ``^{shadow box}'' that is somewhat hard to draw. \TeX\ doesn't balk at ^{negative dimensions}; it just does arithmetic as usual. For example, the combined width of two adjacent boxes is the sum of their widths, whether or not the widths are positive. A font designer can declare a character's width to be negative, in which case the character acts like a ^{backspace}. \ (Languages that read from right to left could be handled in this way, but only to a limited extent, since \TeX's line-breaking ^^{Hebrew} ^^{Arabic} algorithm is based on the assumption that words don't have negative widths.) \danger \TeX\ can raise or lower the individual boxes in a horizontal list; such adjustments take care of mathematical subscripts and superscripts, as well as the heights of accents and a few other things. For example, here is a way to make a box that contains the \TeX\ logo, putting it into \TeX's internal register |\box0|: \begintt \setbox0=\hbox{T\kern-.1667em\lower.5ex\hbox{E}\kern-.125em X} \endtt ^^|\setbox| Here `^|\kern||-.1667em|' means to insert blank space of $-.1667$ ems in the current font, i.e., to back up a bit; and `^|\lower||.5ex|' means that the box |\hbox{E}| is to be lowered by half of the current x-height, thus offsetting that box with respect to the others. Instead of `|\lower.5ex|' one could also say `^|\raise||-.5ex|'. Chapters 12 and~21 discuss the details of how to construct boxes for special effects; our goal in the present chapter is merely to get a taste of the possibilities. \danger \TeX\ will exhibit the contents of any ^{box register}, if you ask it to. For example, if you type `^|\showbox||0|' after setting |\box0| to the \TeX\ logo as above, your ^{log file} will contain the following mumbo jumbo: ^^{TeX logo} \begintt \hbox(6.83331+2.15277)x18.6108 .\tenrm T .\kern -1.66702 .\hbox(6.83331+0.0)x6.80557, shifted 2.15277 ..\tenrm E .\kern -1.25 .\tenrm X \endtt ^^{diagnostic format} ^^{internal box-and-glue representation} ^^{box displays} The first line means that |\box0| is an hbox whose height, depth, and width are respectively $6.83331\pt$, $2.15277\pt$, and $18.6108\pt$. Subsequent lines beginning with `|.|'\ indicate that they are {\sl inside\/} of a box. The first thing in this particular box is the letter~|T| in font |\tenrm|; then comes a kern. The next item is an hbox that contains only the letter~|E|; this box has the height, depth, and width of an |E|, and it has been shifted downward by $2.15277\pt$ (thereby accounting for the depth of the larger box). \dangerexercise Why are there two dots in the `|..\tenrm E|' line here? \answer This |E| is inside a box that's inside a box. \danger Such displays of box contents will be discussed further in Chapters 12 and~17. They are used primarily for diagnostic purposes, when you are trying to figure out exactly what \TeX\ thinks it's doing. The main reason for bringing them up in the present chapter is simply to provide a glimpse of how \TeX\ represents boxes in its guts. A computer program doesn't really move boxes around; it fiddles with lists of representations of boxes. \dangerexercise By running \TeX, figure out how it actually handles italic corrections to characters: How are the corrections represented inside a box? \answer The idea is to construct a box and to look inside. For example, \begintt \setbox0=\hbox{\sl g\/} \showbox0 \endtt reveals that |\/| is implemented by placing a kern after the character. Further experiment shows that this kern is inserted even when the italic correction is zero. \dangerexercise The ``opposite'' of \TeX's logo---namely, T\kern+.1667em\raise.5ex\hbox{E}\kern+.125em X---is produced by \begintt \setbox1=\hbox{T\kern+.1667em\raise.5ex\hbox{E}\kern+.125em X} \endtt What would |\showbox1| show now? \ (Try to guess, without running the machine.) \answer The height, depth, and width of the enclosing box should be just large enough to enclose all of the contents, so the result is: \begintt \hbox(8.98608+0.0)x24.44484 .\tenrm T .\kern 1.66702 .\hbox(6.83331+0.0)x6.80557, shifted -2.15277 ..\tenrm E .\kern 1.25 .\tenrm X \endtt (You probably predicted a height of |8.9861|; \TeX's internal calculations are in |sp|, not |pt|/100000, so the rounding in the fifth decimal place is not readily predictable.) \dangerexercise Why do you think the author of \TeX\ didn't make boxes more symmetrical between horizontal and vertical, by allowing reference points to be inside the boundary instead of insisting that the reference point must appear at the left edge of each box? \answer No applications of such symmetrical boxes to English-language printing were apparent; it seemed pointless to carry extra generality as useless baggage that would rarely if ever be used, merely for the sake of symmetry. In other words, the author wore a computer science cap instead of a mathematician's mantle on the day that \TeX's boxes were born. Time will tell whether or not this was a fundamental error! \ddangerexercise Construct a |\demobox| macro for use in writing manuals like this, so that an author can write `|\demobox{Tough exercise.}|' in order to typeset `\thinspace\demobox{Tough exercise.}\thinspace'. \answer The following solution is based on a general |\makeblankbox| macro that prints the edges of a box using rules of given thickness outside and inside that box; the box dimensions are those of\/ |\box0|.\par |\def\dolist{\afterassignment\dodolist\let\next= }|\parbreak |\def\dodolist{\ifx\next\endlist \let\next\relax|\parbreak | \else \\\let\next\dolist \fi|\parbreak | \next}|\par |\def\endlist{\endlist}|\par |\def\hidehrule#1#2{\kern-#1%|\parbreak | \hrule height#1 depth#2 \kern-#2 }|\par |\def\hidevrule#1#2{\kern-#1{\dimen0=#1|\parbreak | \advance\dimen0 by#2\vrule width\dimen0}\kern-#2 }|\par |\def\makeblankbox#1#2{\hbox{\lower\dp0\vbox{\hidehrule{#1}{#2}%|\parbreak | \kern-#1 % overlap the rules at the corners|\parbreak | \hbox to \wd0{\hidevrule{#1}{#2}%|\parbreak | \raise\ht0\vbox to #1{}% set the vrule height|\parbreak | \lower\dp0\vtop to #1{}% set the vrule depth|\parbreak | \hfil\hidevrule{#2}{#1}}%|\parbreak | \kern-#1\hidehrule{#2}{#1}}}}|\par |\def\maketypebox{\makeblankbox{0pt}{1pt}}|\par |\def\makelightbox{\makeblankbox{.2pt}{.2pt}}|\par |\def\\{\if\space\next\ % assume that \next is unexpandable|\parbreak | \else \setbox0=\hbox{\next}\maketypebox\fi}|\par |\def\demobox#1{\setbox0=\hbox{\dolist#1\endlist}%|\parbreak | \leavevmode\copy0\kern-\wd0\makelightbox}|\par \def\frac#1/#2{\leavevmode\kern.1em \raise.5ex\hbox{\the\scriptfont0 #1}\kern-.1em /\kern-.15em\lower.25ex\hbox{\the\scriptfont0 #2}} \ddangerexercise Construct a |\frac| macro such that `|\frac1/2|' yields `\frac1/2'. \checkequals\fracexno\exno \answer |\def\frac#1/#2{\leavevmode\kern.1em|\parbreak |\raise.5ex\hbox{\the\scriptfont0 #1}\kern-.1em|\parbreak |/\kern-.15em\lower.25ex\hbox{\the\scriptfont0 #2}}| \endchapter I have several boxes in my memory in which I will keep them all very safe, % he's talking about "instructions" there shall not a one of them be lost. \author IZAAK ^{WALTON}, {\sl The Compleat Angler\/} (1653) % beginning Chap12 % in 1654 and subsequent editions, this quote comes in Chap17 % the 1653 spelling agrees with 20th century conventions in this passage! \bigskip How very little does the amateur, dwelling at home at ease, comprehend the labours and perils of the author. \author R. L. ^{STEVENSON} and L. ^{OSBOURNE}, {\sl The Wrong Box\/} (1889) \eject \beginchapter Chapter 12. Glue But there's more to the story than just boxes: There's also some magic mortar called {\sl ^{glue}\/} that \TeX\ uses to paste boxes together. For example, there is a little space between the lines of text in this manual; it has been calculated so that the baselines of consecutive lines within a paragraph are exactly 12~points apart. And there is space between words too; such space is not an ``empty'' box, it is part of the glue between boxes. This glue can stretch or shrink so that the right-hand margin of each page comes out looking straight. ^^{leading, see baselineskip} ^^{skipping space, see glue} When \TeX\ makes a large box from a horizontal or vertical list of smaller boxes, there often is glue between the smaller boxes. Glue has three attributes, namely its natural {\sl space}, its ability to {\sl ^{stretch}}, and its ability to {\sl ^{shrink}}. In order to understand how this works, consider the following example of four boxes in a horizontal list separated by three globs of glue: \begindisplay\eightpoint \vbox{ \hbox{\samplebox{7mm}{8mm}{5\varunit}{width 5}% \sampleglue{9\varunit}{space 9\cr stretch 3\cr shrink 1}% \samplebox{3mm}{2mm}{6\varunit}{width 6}% \sampleglue{9\varunit}{space 9\cr stretch 6\cr shrink 2}% \samplebox{8mm}{3mm}{3\varunit}{width 3}% \sampleglue{12\varunit}{space 12\cr stretch 0\cr shrink 0}% \samplebox{4mm}{7mm}{8\varunit}{width 8}} \kern6pt \arrows{52\varunit}{width 52}} \enddisplay The first glue element has 9 units of space, 3 of stretch, and 1 of shrink; the next one also has 9 units of space, but 6 units of stretch and 2 of shrink; the last one has 12 units of space, but it is unable to stretch or to shrink, so it will remain 12 units of space no matter what. The total width of boxes and glue in this example, considering only the space components of the glue, is $5+9+6+9+3+12+8=52$ units. This is called the {\sl ^{natural width}\/} of the horizontal list; it's the preferred way to paste the boxes together. Suppose, however, that \TeX\ is told to make the horizontal list into a box that is 58~units wide; then the glue has to stretch by 6~units. Well, there are $3+6+0=9$ units of stretchability present, so \TeX\ multiplies each unit of stretchability by 6/9 in order to obtain the extra 6~units needed. The first glob of glue becomes $9+(6/9)\times3=11$ units wide, the next becomes $9+(6/9)\times6=13$ units wide, the last remains 12 units wide, and we obtain the desired box looking like this: \begindisplay\eightpoint \vbox{\kern-3pt \hbox{\samplebox{7mm}{8mm}{5\varunit}{}% \sampleglue{11\varunit}{$9+2$}% \samplebox{3mm}{2mm}{6\varunit}{}% \sampleglue{13\varunit}{$9+4$}% \samplebox{8mm}{3mm}{3\varunit}{}% \sampleglue{12\varunit}{$12+0$}% \samplebox{4mm}{7mm}{8\varunit}{}} \kern6pt \arrows{58\varunit}{width 58}} \enddisplay On the other hand, if \TeX\ is supposed to make a box 51 units wide from the given list, it is necessary for the glue to shrink by a total of one unit. There are three units of shrinkability present, so the first glob of glue would shrink by 1/3 and the second by 2/3. \smallbreak The process of determining glue thickness when a box is being made from a horizontal or vertical list is called {\sl ^{setting the glue}}. Once glue has been set, it becomes rigid; it won't stretch or shrink any more, and the resulting box is essentially indecomposable. Glue will never shrink more than its stated shrinkability. For example, the first glob of glue in our illustration will never be allowed to become narrower than 8 units wide, and \TeX\ will never shrink the given horizontal list to make its total width less than 49 units. But glue is allowed to stretch arbitrarily far, whenever it has a positive stretch component. \exercise How wide would the glue globs be if the horizontal list in the illustration were to be made 100 units wide? \answer $9+16$ units, $9+32$ units, $12+0$ units. \ (But \TeX\ would consider so much stretching to be ``infinitely bad.'') Once you understand \TeX's concept of glue, you may well decide that it was misnamed; real glue doesn't stretch or shrink in such ways, nor does it contribute much space between boxes that it welds together. Another word like ``spring'' would be much closer to the essential idea, since ^{springs} have a natural width, and since different springs compress and expand at different rates under tension. But whenever the author has suggested changing \TeX's terminology, numerous people have said that they like the word ``glue'' in spite of its inappropriateness; so the original name has stuck. \danger \TeX\ is somewhat reluctant to stretch glue more than the stated stretchability; therefore you can decide how big to make each aspect of the glue by using the following rules: \ (a)~The natural glue space should be the amount of space that looks best. \ (b)~The glue stretch should be the maximum amount of space that can be added to the natural spacing before the layout begins to look bad. \ (c)~The glue shrink should be the maximum amount of space that can be subtracted from the natural spacing before the layout begins to look bad. In most cases the designer of a book layout will have specified all the kinds of glue that are to be used, so a typist will not need to decide how big any glue attributes should be. For example, users of the plain \TeX\ format of Appendix~B can type `|\smallskip|' when they want a little extra ^{space between paragraphs}; a ^|\smallskip| turns out to be $3\pt$ worth of vertical glue that can stretch or shrink by an additional~$1\pt$. Here is a |\smallskip|: \smallskip \noindent Instead of sprinkling various amounts of glue throughout a manuscript, expressing each of them explicitly in terms of points, you will find it much better to explain your intentions more clearly by typing something like `|\smallskip|' when you want abnormal spacing. The definition of\/ |\smallskip| can readily be changed later, in case you want such spaces to be smaller or larger. Plain \TeX\ also provides you with `^|\medskip|', which is worth two smallskips, and `^|\bigskip|', which is worth two medskips. \danger A plain \TeX\ |\medskip| appears before and after each ``^{dangerous bend}'' section of this manual, so you have already seen numerous examples of such spacing before you knew what it was called. Vertical glue is created by writing `|\vskip|\', where ^\ is any glue specification. The usual way to specify \ to \TeX\ is \begindisplay \ |plus|\ |minus|\ \enddisplay where the `|plus|\' and `|minus|^\' are optional and assumed to be zero if not present; `^|plus|' introduces the amount of stretchability, `^|minus|' introduces the amount of shrinkability. For example, Appendix~B defines |\medskip| to be an abbreviation for `|\vskip6pt plus2pt minus2pt|'. The normal-space component of glue must always be given as an explicit \, even when it is zero. \danger Horizontal glue is created in the same way, but with ^|\hskip| instead of\/ ^|\vskip|. For example, plain \TeX\ defines ^|\enskip| as an abbreviation for the command `|\hskip.5em\relax|'; this skips horizontally by one ``^{en},'' i.e., by exactly half of an em in the current font. There is no stretching or shrinking in an |\enskip|. The control sequence ^|\relax| after `|.5em|' prevents \TeX\ from thinking that a ^{keyword} is present, in case the text following |\enskip| just happens to begin with `|plus|' or `|minus|'. One of the interesting things that can happen when glue stretches and shrinks at different rates is that there might be glue with {\sl ^{infinite}\/} stretchability. For example, consider again the four boxes we had at the beginning of this chapter, with the same glue as before except that the glue in the middle can stretch infinitely far. Now the total stretchability is infinite; and when the line has to grow, all of the additional space is put into the middle glue. If, for example, a box of width 58 is desired, the middle glue expands from 9 to~15 units, and the other spacing remains unchanged. If such infinitely stretchable glue is placed at the left of a row of boxes, the effect is to place them ``flush right,'' i.e., to move them over to the ^^{right justification} ^^{centering} ^^{flush right} rightmost boundary of the constructed box. And if you take {\sl two\/} globs of infinitely stretchable glue, putting one at the left and one at the right, the effect is to {\sl center\/} the list of boxes within a larger box. This in fact is how the ^|\centerline| instruction works in plain \TeX\null: It places infinite glue at both ends, then makes a box whose width is the current value of\/ |\hsize|. The short story example of Chapter 6 used infinite glue not only for centering, but also in the ^|\vfill| instruction at the end; `|\vfill|' essentially means ``skip vertically by zero, but with infinite stretchability.'' In other words, |\vfill| fills up the rest of the current page with blank space. \danger \TeX\ actually recognizes several kinds of infinity, some of which are ``more infinite'' than others. You can say both ^|\vfil| and |\vfill|; the second is stronger than the first. In other words, if no other infinite stretchability is present, |\vfil| will expand to fill the remaining space; but if both |\vfil| and |\vfill| are present simultaneously, the |\vfill| effectively prevents |\vfil| from stretching. You can think of it as if\/ |\vfil| has one mile of stretchability, while |\vfill| has a trillion miles. \danger Besides |\vfil| and |\vfill|, \TeX\ has ^|\hfil| and ^|\hfill|, for stretching indefinitely in the horizontal direction. You can also say ^|\hss| or ^|\vss|, in order to get glue that is infinitely shrinkable as well as infinitely stretchable. \ (The name `|\hss|' stands for ``horizontal stretch or shrink''; `|\vss|' is its vertical counterpart.) \ Finally, the primitives ^|\hfilneg| and ^|\vfilneg| will cancel the stretchability of |\hfil| and |\vfil|; we shall discuss applications of these curious glues later. \danger Here are some examples of\/ |\hfil|, using the ^|\line| macro of plain \TeX, which creates an hbox whose width is the current |\hsize|: ^^{flush left} \begintt \line{This text will be flush left.\hfil} \line{\hfil This text will be flush right.} \line{\hfil This text will be centered.\hfil} \line{Some text flush left\hfil and some flush right.} \line{Alpha\hfil centered between Alpha and Omega\hfil Omega} \line{Five\hfil words\hfil equally\hfil spaced\hfil out.} \endtt \dangerexercise Describe the result of \xdef\linexno{\the\exno}% \begintt \line{\hfil\hfil What happens now?\hfil} \line{\hfill\hfil and now?\hfil} \endtt \answer `What happens now?' is placed in a line of width |\hsize|, with twice as much space at the left as at the right; `and now?' is put flush right on the following line. \ddangerexercise How do the following three macros behave differently? \begintt \def\centerlinea#1{\line{\hfil#1\hfil}} \def\centerlineb#1{\line{\hfill#1\hfill}} \def\centerlinec#1{\line{\hss#1\hss}} \endtt \answer The first two give an ``overfull box'' if the argument doesn't fit on a line; the third allows the argument to stick out into the margins instead. \ (Plain \TeX's ^|\centerline| is |\centerlinec|; the stickout effect shows up in the narrow-column experiment of Chapter~6.) \ If the argument contains no infinite glue, |\centerlinea| and |\centerlineb| produce the same effect; but |\centerlineb| will center an argument that contains `fil' glue. \danger In order to specify such infinities, you are allowed to use the special units `^|fil|', `^|fill|', and `^|filll|' in the \ parts of a stretchability or shrinkability component. For example, |\vfil|, |\vfill|, |\vss|, and |\vfilneg| are essentially equivalent to the glue specifications \begintt \vskip 0pt plus 1fil \vskip 0pt plus 1fill \vskip 0pt plus 1fil minus 1fil \vskip 0pt plus -1fil \endtt respectively. It's usually best to stick to the first order infinity (fil) as much as you can, resorting to second order (fill) only when you really need something extremely infinite. Then the ultimate order (filll) is always available as a last resort in emergencies. \ (\TeX\ does not provide a `^|\vfilll|' primitive, since the use of this highest infinity is not encouraged.) \ You can use fractional multiples of infinity like `|3.25fil|', as long as you stick to fewer than 16384 fil units. \TeX\ actually does its calculations with integer multiples of $\rm2^{-16}\,fil$ (or fill or filll); so |0.000007filll| turns out to be indistinguishable from |0pt|, but |0.00001filll| is infinitely greater than |16383.99999fill|. Now here's something important for all \TeX nical typists to know: Plain \TeX\ puts extra space at the end of a ^{sentence}; furthermore, it automatically increases the stretchability (and decreases the shrinkability) after ^{punctuation} marks. The reason is that it's usually better to put more space after punctuation than between two ordinary words, when spreading a line out to reach the desired margins. Consider, for example, the following sentences from a classic kindergarten pre-primer: ^^{Dick and Jane} \begintt ``Oh, oh!'' cried Baby Sally. Dick and Jane laughed. \endtt If \TeX\ sets this at its natural width, all the spaces will be the same, except after the quote and after `Baby Sally.': \begindisplay ``Oh, oh!'' cried Baby Sally. Dick and Jane laughed. \enddisplay But if the line needs to be expanded by 5 points, 10 points, 15 points, or more, \TeX\ will set it as \begindisplay \hbox spread 5pt{``Oh, oh!'' cried Baby Sally. Dick and Jane laughed.}\cr \hbox spread 10pt{``Oh, oh!'' cried Baby Sally. Dick and Jane laughed.}\cr \hbox spread 15pt{``Oh, oh!'' cried Baby Sally. Dick and Jane laughed.}\cr \hbox spread 20pt{``Oh, oh!'' cried Baby Sally. Dick and Jane laughed.}\cr \enddisplay The glue after the comma stretches at 1.25 times the rate of the glue between adjacent words; the glue after the period and after the |!''| stretches at 3 times the rate. There is no glue between adjacent letters, so individual words will always look the same. If \TeX\ had to shrink this line to its minimum width, the result would be \begindisplay \hfuzz 1000pt \hbox to 0pt{``Oh, oh!'' cried Baby Sally. Dick and Jane laughed.} \enddisplay The glue after a ^{comma} shrinks only 80 percent as much as ordinary inter-word glue, and after a ^{period} or ^{exclamation point} or ^{question mark} it shrinks by only one third as much. This all makes for nice-looking output, but it unfortunately adds a bit of a burden to your job as a typist, because \TeX's rule for determining the end of a sentence {\sl doesn't always work}. The problem is that a period sometimes comes in the middle of a sentence $\ldots$ like when it is used (as here) to make an ``^{ellipsis}'' of three dots. ^^{three dots, see ellipsis} ^^{dot dot dot, see ellipsis} Moreover, if you try to specify `$\ldots$' by typing three periods in a row, you get `...'---the dots are too close together. One way to handle this is to go into {\sl mathematics\/} mode, using the ^|\ldots| control sequence defined in plain \TeX\ format. For example, if you type \begintt Hmmm $\ldots$ I wonder why? \endtt the result is `Hmmm $\ldots$ I wonder why?'. This works because math formulas are exempt from the normal text spacing rules. Chapter 18 has more to say about |\ldots| and related topics. ^{Abbreviations} present problems too. For example, the short story in Chapter~6 referred to `Mr.~^{Drofnats}'; \TeX\ must be told somehow that the period after `Mr.'\ or `Mrs.'\ or `Ms.' or `Prof.'\ or `Dr.' or `Rt.~Hon.', etc., doesn't count as a sentence-ending ^{full stop}. We avoided that embarrassment in Chapter~6 by typing `|Mr.~Drofnats|'; the ``^{tie}'' mark |~| ^^{tilde} tells plain \TeX\ to insert a normal space, and to refrain from breaking between lines at that space. Another way to get \TeX\ to put out a normal space is to type `|\|\]' (^{control space}); e.g., `|Mr.\ Drofnats|' would be almost the same as `|Mr.~Drofnats|', except that a line might end after the `Mr.'. The tie mark is best for abbreviations within a name, and after several other common abbreviations like `Fig.'\ and `cf.'\ and `vs.'\ and `resp.'; you will find that it's easy to train yourself to type `|cf.~Fig.~5|'. In fact, it's usually wise to type |~| (instead of a space) just after a common abbreviation that occurs in the middle of a sentence. Manuals of style will tell you that the abbreviations `e.g.'\ and `i.e.'\ should always be followed by commas, never by spaces, so those particular cases shouldn't need any special treatment. The only remaining abbreviations that arise with significant frequency occur in bibliographic references; ^{control spaces} are appropriate here. ^^{interword spacing} If, for example, you are typing a manuscript that refers to `Proc.\ Amer.\ Math.\ Soc.', you should say \begintt Proc.\ Amer.\ Math.\ Soc. \endtt Granted that this input looks a bit ugly, it makes the output look right. It's one of the things we occasionally must do when dealing with a computer that tries to be smart. \exercise Explain how to type the following sentence: ``Mr.~\& Mrs.~^{User} were married by Rev.~^{Drofnats}, who preached on Matt.~19\thinspace:\thinspace3--9.'' \answer |Mr.~\& Mrs.~User were married by Rev.~Drofnats, who preached on Matt.~19\thinspace:\thinspace3--9.| \ (Such thin spaces are traditional for ^{Biblical references} to chapter and verse, but you weren't really expected to know that. Plain \TeX\ defines ^|\thinspace| to be a kern, not glue; hence no break between lines will occur at a thinspace.) \exercise Put the following bibliographic reference into plain \TeX\ language: Donald~E. ^{Knuth}, ``Mathematical typography,'' {\sl Bull.\ Amer.\ Math.\ Soc.\ \bf1} (1979), 337--372. \answer |Donald~E.\ Knuth, ``Mathematical typography,'' {\sl Bull.\ Amer.\ Math.\ Soc.\ \bf1} (1979), 337--372.| \ (But the `|\|' after `|E.|' isn't necessary, because of a rule you will learn if you venture around the next dangerous bend.) On the other hand, if you don't care about such refinements of spacing you can tell plain \TeX\ to make all spaces the same, regardless of punctuation marks, by simply typing `^|\frenchspacing|' at the beginning of your manuscript. French spacing looks like this: \begindisplay \frenchspacing ``Oh, oh!'' cried Baby Sally. Dick and Jane laughed. \enddisplay You can also shift back and forth between the two styles, either by saying `^|\nonfrenchspacing|' to establish ^{sophisticated spacing}, or by making your use of\/ |\frenchspacing| local to some group. For example, you might want to use French spacing only when typing the bibliography of some document. \danger \TeX\ doesn't consider a period or question mark or exclamation point to be the end of a sentence if the preceding character is an uppercase letter, since \TeX\ assumes that such uppercase letters are most likely somebody's initials. Thus, for example, the `|\|' is unnecessary after the~`|I.|' in `|Dr.~Livingstone~I.\ Presume|'; that particular period is not assumed to be a full stop. ^^{Presume} \dangerexercise What can you do to make \TeX\ recognize the ends of sentences that do end with uppercase letters (e.g., `$\ldots$ launched by NASA.\null' or `Did I?' or `$\ldots$ see Appendix~A.')? \answer There are several ways; perhaps the easiest are to type `|\hbox{NASA}.|'\ or `|NASA\null.|' \ (The ^|\null| macro is an abbreviation for `|\hbox{}|'.) \danger You can see the glue that \TeX\ puts between words by looking at the contents of hboxes in the internal ^{diagnostic format} that we discussed ^^{internal box-and-glue representation} briefly in Chapter~11. For example, Baby Sally's exclamation begins as follows, after \TeX\ has digested it and put it into a box, assuming |\nonfrenchspacing|: \begintt .\tenrm \ (ligature ``) .\tenrm O .\tenrm h .\tenrm , .\glue 3.33333 plus 2.08331 minus 0.88889 .\tenrm o .\tenrm h .\tenrm ! .\tenrm " (ligature '') .\glue 4.44444 plus 4.99997 minus 0.37036 .\tenrm c .\tenrm r .\tenrm i .\tenrm e .\tenrm d .\glue 3.33333 plus 1.66666 minus 1.11111 .\tenrm B .\tenrm a .\tenrm b .\kern-0.27779 .\tenrm y .\glue 3.33333 plus 1.66666 minus 1.11111 .\tenrm S .\tenrm a .\tenrm l .\tenrm l .\tenrm y .\kern-0.83334 .\tenrm . .\glue 4.44444 plus 4.99997 minus 0.37036 \endtt The normal ^{interword glue} in font |\tenrm| is $3.33333\pt$, plus $1.66666\pt$ of stretchability, minus $1.11111\pt$ of shrinkability. Notice that the interword |\glue| in this list stretches more, and shrinks less, after the punctuation marks; and the natural space is in fact larger at the end of each sentence. This example also shows several other things that \TeX\ does while it processes the sample line of text: It converts |``| and |''| into single characters, i.e., ^{ligatures}; and it inserts small ^{kerns} in two places to improve the spacing. A ^|\kern| is similar to glue, but it is not the same, because kerns cannot stretch or shrink; furthermore, \TeX\ will never break a line at a kern, unless that kern is immediately followed by glue. \ddanger You may be wondering what \TeX's rules for interword glue really are, exactly. For example, how did \TeX\ remember the effect of Baby Sally's exclamation point, when quotation marks intervened before the next space? The details are slightly tricky, but not incomprehensible. When \TeX\ is processing a horizontal list of boxes and glue, it keeps track of a positive integer called the current ``^{space factor}.'' The space factor is normally 1000, which means that the interword glue should not be modified. If the space factor $f$ is different from 1000, the interword glue is computed as follows: Take the normal space glue for the current font, and add the extra space if $f\ge2000$. \ (Each font specifies a normal space, normal stretch, normal shrink, and extra space; for example, these quantities are $3.33333\pt$, $1.66666\pt$, $1.11111\pt$, and $1.11111\pt$, respectively, in ^|cmr10|. We'll discuss such font parameters in greater detail later.) \ ^^|\fontdimen| Then the stretch component is multiplied by $f/1000$, while the shrink component is multiplied by $1000/f$. \ddanger However, \TeX\ has two parameters ^|\spaceskip| and ^|\xspaceskip| that allow you to override the normal spacing of the current font. If $f\ge2000$ and if\/ |\xspaceskip| is nonzero, the |\xspaceskip| glue is used for an ^{interword space}. Otherwise if\/ |\spaceskip| is nonzero, the |\spaceskip| glue is used, with stretch and shrink components multiplied by $f/1000$ and $1000/f$. For example, the ^|\raggedright| macro of plain \TeX\ uses |\spaceskip| and |\xspaceskip| to suppress all stretching and shrinking of interword spaces. \ddanger The space factor $f$ is 1000 at the beginning of a horizontal list, and it is set to 1000 just after a non-character box or a math formula has been put onto the current horizontal list. You can say `^|\spacefactor||=|\' to assign any particular value to the space factor; but ordinarily, $f$ gets set to a number other than 1000 only when a simple character box goes on the list. Each character has a ^{space factor code}, and when a character whose space factor code is $g$ enters the current list the normal procedure is simply to assign $g$ as the new space factor. However, if $g$ is zero, $f$ is not changed; and if $f<1000w$, \TeX\ attempts to shrink the contents of the box in a similar way; the glue order is the highest subscript~$i$ such that $z_i\ne0$, and the glue ratio is normally $r=(x-w)/z_i$. However, $r$ is set to 1.0 in the case $i=0$ and $x-w>z_0$, because the maximum shrinkability must not be exceeded. \ (c)~Finally, every glob of glue in the horizontal list being boxed is modified. Suppose the glue has natural width~$u$, stretchability~$y$, and shrinkability~$z$, where $y$~is a $j$th order infinity and $z$~is a $k$th order infinity. Then if $xw$ (shrinking), this glue takes the new width $u-rz$ if $k=i$; it keeps its natural width~$u$ if $k\ne i$. Notice that stretching or shrinking occurs only when the glue has the highest order of infinity that doesn't cancel out. \danger \TeX\ will construct an hbox that has a given width $w$ if you issue the command `\hbox{|\hbox to |\|{|\|}|}', where $w$ is the value of the \. For example, the ^|\line| macro discussed earlier in this chapter is simply an abbreviation for `|\hbox to\hsize|'. ^^|to|^^|\hbox| \TeX\ also allows you to specify the exact amount of stretching or shrinking; the command `\hbox{|\hbox spread|\|{|\|}|}' creates a box whose width~$w$ is a given amount more ^^|spread| than the natural width of the contents. For example, one of the boxes displayed earlier in this chapter was generated by \begintt \hbox spread 5pt{``Oh, oh!'' ... laughed.} \endtt In the simplest case, when you just want a box to have its natural width, you don't have to write `|\hbox spread 0pt|'; you can simply say `|\hbox{|\|}|'. \danger The ^{baseline} of a constructed hbox is the common baseline of the boxes inside. \ (More precisely, it's the common baseline that they would share if they weren't raised or lowered.) \ The height and depth of a constructed hbox are determined by the maximum distances by which the interior boxes reach above and below the baseline, respectively. The result of\/ |\hbox| never has negative height or negative depth, but the width can be negative. \dangerexercise Assume that |\box1| is $1\pt$~high, $1\pt$~deep, and $1\pt$~wide; |\box2| is $2\pt$~high, $2\pt$~deep, and $2\pt$~wide. A third box is formed by saying ^^|\setbox| \begintt \setbox3=\hbox to3pt{\hfil\lower3pt\box1\hskip-3pt plus3fil\box2} \endtt What are the height, depth, and width of\/ |\box3|? Describe the position of the reference points of boxes 1 and~2 with respect to the reference point of box~3. \answer |\box3| is $2\pt$ high, $4\pt$ deep, $3\pt$ wide. Starting at the reference point of\/ |\box3|, go right $.75\pt$ and down $3\pt$ to reach the reference point of\/ |\box1|; or go right $1\pt$ to reach the reference point of\/ |\box2|. \danger The process of setting glue for vboxes is similar to that for hboxes; but before we study the |\vbox| operation, we need to discuss how \TeX\ stacks boxes up vertically so that their baselines tend to be a fixed distance apart. The boxes in a horizontal list often touch each other, but it's usually wrong to do this in a vertical list; imagine how awful a page would look if its lines of type were brought closer together whenever they didn't contain tall letters, or whenever they didn't contain any letters that descended below the baseline. \danger \TeX's solution to this problem involves three primitives called ^|\baselineskip|, ^|\lineskip|, and ^|\lineskiplimit|. A format designer chooses values of these three quantities by writing \begindisplay |\baselineskip=|\\cr |\lineskip=|\\cr |\lineskiplimit=|\\cr \enddisplay and the interpretation is essentially this: Whenever a box is added to a vertical list, \TeX\ inserts ``^{interline glue}'' intended to make the distance between the baseline of the new box and the baseline of the previous box exactly equal to the value of\/ |\baselineskip|. But if the interline glue calculated by this rule would cause the top edge of the new box to be closer than |\lineskiplimit| to the bottom edge of the previous box, then |\lineskip| is used as the interline glue. In other words, the distance between adjacent baselines will be the |\baselineskip| setting, unless that would bring the boxes too close together; the |\lineskip| glue will separate adjacent boxes in the latter case. \danger The rules for interline glue in the previous paragraph are carried out without regard to other kinds of glue that might be present; all vertical spacing due to explicit appearances of\/ |\vskip| and |\kern| acts independently of the interline glue. Thus, for example, a ^|\smallskip| between two lines always makes their baselines further apart than usual, by the amount of a |\smallskip|; it does not affect the decision about whether |\lineskip| glue is used between those lines. \danger For example, let's suppose that |\baselineskip=12pt plus 2pt|, |\lineskip=|\penalty0|3pt minus 1pt|, and |\lineskiplimit=2pt|. \ (These values aren't particularly useful; they have simply been chosen to illustrate the rules.) \ Suppose further that a box whose depth is $3\pt$ was most recently added to the current vertical list; we are about to add a new box whose height is~$h$. If $h=5\pt$, the interline glue will be $4\pt$~plus~$2\pt$, since this will make the baselines $12\pt$~plus~$2\pt$ apart when we add $h$ and the previous depth to the interline glue. But if $h=8\pt$, the interline glue will be $3\pt$~minus~$1\pt$, since |\lineskip| will be chosen in order to keep from violating the given |\lineskiplimit| when stretching and shrinking are ignored. \danger When you are typesetting a document that spans several pages, it's generally best to define the |\baselineskip| so that it cannot stretch or shrink, because this will give more uniformity to the pages. A small variation in the distance between baselines---say only half a point---can make a substantial difference in the appearance of the type, since it significantly affects the proportion of white to black. On the other hand, if you are preparing a one-page document, you might want to give the baselineskip some stretchability, so that \TeX\ will help you fit the copy on the page. \dangerexercise What settings of\/ |\baselineskip|, |\lineskip|, and |\lineskiplimit| will cause the interline glue to be a ``continuous'' function of the next box height (i.e., the interline glue will never change a lot when the box height changes only a little)? \answer The stretch and shrink components of\/ |\baselineskip| and |\lineskip| should be equal, and the |\lineskiplimit| should equal the normal |\lineskip| spacing, to guarantee continuity. \danger A study of \TeX's ^{internal box-and-glue representation} should help to firm up some of these ideas. Here is an excerpt from the vertical list that \TeX\ constructed when it was typesetting this very paragraph: \begintt \glue 6.0 plus 2.0 minus 2.0 \glue(\parskip) 0.0 plus 1.0 \glue(\baselineskip) 1.25 \hbox(7.5+1.93748)x312.0, glue set 0.80154, shifted 36.0 [] \penalty 10000 \glue(\baselineskip) 2.81252 \hbox(6.25+1.93748)x312.0, glue set 0.5816, shifted 36.0 [] \penalty 50 \glue(\baselineskip) 2.81252 \hbox(6.25+1.75)x348.0, glue set 116.70227fil [] \penalty 10000 \glue(\abovedisplayskip) 6.0 plus 3.0 minus 1.0 \glue(\lineskip) 1.0 \hbox(149.25+0.74998)x348.0 [] \endtt {\showboxdepth0\showboxbreadth9999\batchmode\showlists\errorstopmode}% % The log file now contains lines like these; I copied them % into this MS! But I deleted an `insert' for the index... The first |\glue| in this example is the ^|\medskip| that precedes each dangerous-bend paragraph. Then comes the ^|\parskip| glue, which is automatically supplied before the first line of a new paragraph. Then comes some interline glue of $1.25\pt$; it was calculated to make a total of $11\pt$ when the height of the next box ($7.5\pt$) and the depth of the previous box were added. \ (The previous box is not shown---it's the bottom line of exercise~\chapno.\the\exno---but we can deduce that its depth was $2.25\pt$.) \ The |\hbox| that follows is the first line of this paragraph; it has been shifted right $36\pt$ because of ^{hanging indentation}. The glue set ratio for this hbox is 0.80154; i.e., the glue inside is stretched by 80.154\%\ of its stretchability. \ (In the case of shrinking, the ratio following `^|glue set|' would have been preceded by `|- |'; hence we know that stretching is involved here.) \ \TeX\ has put `|[]|' at the end of each hbox line to indicate that there's something in the box that isn't shown. \ (The box contents would have been displayed completely, if ^|\showboxdepth| had been set higher.) \ The ^|\penalty| indications are used to discourage bad breaks between pages, as we will see later. The third hbox has a glue ratio of 116.70227, which applies to first-order-infinite stretching (i.e., fil); this results from an |\hfil| that was implicitly inserted just before the displayed material, to fill up the third line of the paragraph. Finally the big hbox whose height is $149.25\pt$ causes |\lineskip| to be the interline glue. This large box contains the individual lines of typewriter type that are displayed; they have been packaged into a single box so that they cannot be split between pages. Careful study of this example will teach you a lot about \TeX's inner workings. \danger Exception: No interline glue is inserted before or after a rule box. You can also inhibit interline glue by saying ^|\nointerlineskip| between boxes. \ddanger \TeX's implementation of interline glue involves another primitive quantity called ^|\prevdepth|, which usually contains the depth of the most recent box on the current vertical list. However, |\prevdepth| is set to the sentinel value $-1000\pt$ at the beginning of a vertical list, or just after a rule box; this serves to suppress the next interline glue. The user can change the value of\/ |\prevdepth| at any time when building a vertical list; thus, for example, the |\nointerlineskip| macro of Appendix~B simply expands to `|\prevdepth=-1000pt|'. \ddanger Here are the exact rules by which \TeX\ calculates the interline glue between boxes: Assume that a new box of height~$h$ (not a rule box) is about to be appended to the bottom of the current vertical list, and let $\hbox{|\prevdepth|}=p$, $\hbox{|\lineskiplimit|}=l$, \hbox{$\hbox{|\baselineskip|}=(b$ plus~$y$ minus~$z)$}. If $p\le-1000\pt$, no interline glue is added. Otherwise if $b-p-h\ge l$, the interline glue `$(b-p-h)$ plus~$y$ minus~$z$' will be appended just above the new box. Otherwise the |\lineskip| glue will be appended. Finally, |\prevdepth| is set to the depth of the new box. \ddangerexercise Mr.~B. L. ^{User} had an application in which he wanted to put a number of boxes together in a vertical list, with no space between them. He didn't want to say |\nointerlineskip| after each box; so he decided to set |\baselineskip|, |\lineskip|, and |\lineskiplimit| all equal to |0pt|. Did this work? \answer Yes it did, but only because none of his boxes had a negative height or depth. He would have been safer if he had set |\baselineskip=-1000pt|, |\lineskip=0pt|, and |\lineskiplimit=16383pt|. \ (Plain \TeX's ^|\offinterlineskip| macro does this.) \danger The vertical analog of\/ |\hbox| is ^|\vbox|, and \TeX\ will obey the commands `|\vbox to|\' and `|\vbox spread|\' in about the way you would expect, by analogy with the horizontal case. However, there's a slight complication because boxes have both height and depth in the vertical direction, while they have only width in the horizontal direction. The dimension in a |\vbox| command refers to the final height of the vbox, so that, for example, `|\vbox to 50pt{...}|' produces a box that is $50\pt$ high; this is appropriate because everything that can stretch or shrink inside a vbox appears in the part that contributes to the height, while the depth is unaffected by glue setting. \danger The depth of a constructed |\vbox| is best thought of as the depth ^^{depth of box} ^^{height of box} of the bottom box inside. Thus, a vbox is conceptually built by taking a bunch of boxes and arranging them so that their reference points are lined up vertically; then the reference point of the lowest box is taken as the reference point of the whole, and the glue is set so that the final height has some desired value. \danger However, this description of vboxes glosses over some technicalities that come up when you consider unusual cases. For example, \TeX\ allows you to shift boxes in a vertical list to the right or to the left by saying `^|\moveright|\\' or `^|\moveleft|\\'; this is like the ability to ^|\raise| or ^|\lower| boxes in a horizontal list, and it implies that the reference points inside a vbox need not always lie in a vertical line. Furthermore, it is necessary to guard against boxes that have too much depth, lest they extend too far into the bottom margin of a page; and later chapters will point out that vertical lists can contain other things like penalties and marks, in addition to boxes and glue. \ddanger Therefore, the actual rules for the depth of a constructed vbox are somewhat \TeX nical. Here they are: Given a vertical list that is being wrapped up via |\vbox|, the problem is to determine its natural depth. \ (1)~If the vertical list contains no boxes, the depth is zero. \ (2)~If there's at least one box, but if the final box is followed by kerning or glue, possibly with intervening penalties or other things, the depth is zero. \ (3)~If there's at least one box, and if the final box is not followed by kerning or glue, the depth is the depth of that box. \ (4)~However, if the depth computed by rules (1), (2), or~(3) exceeds ^|\boxmaxdepth|, the depth will be the current value of\/ |\boxmaxdepth|. \ (Plain \TeX\ sets |\boxmaxdepth| to the largest possible dimension; therefore rule~(4) won't apply unless you specify a smaller value. When rule~(4) does decrease the depth, \TeX\ adds the excess depth to the box's natural height, essentially moving the reference point down until the depth has been reduced to the stated maximum.) \danger The glue is set in a vbox just as in an hbox, by determining a ^{glue set ratio} and a ^{glue set order}, based on the difference between the natural height~$x$ and the desired height~$w$, and based on the amounts of stretchability and shrinkability that happen to be present. \danger The width of a computed |\vbox| is the maximum distance by which an enclosed box extends to the right of the reference point, taking possible shifting into account. This width is always nonnegative. \dangerexercise Assume that |\box1| is $1\pt$ high, $1\pt$ deep, and $1\pt$ wide; |\box2| is $2\pt$ high, $2\pt$ deep, and $2\pt$ wide; the baselineskip, lineskip, and lineskiplimit are all zero; and the |\boxmaxdepth| is very large. A third box is formed by saying \begintt \setbox3=\vbox to3pt{\moveright3pt\box1\vskip-3pt plus3fil\box2} \endtt What are the height, depth, and width of\/ |\box3|? Describe the position of the reference points of boxes 1 and~2 with respect to the reference point of box~3. \answer The interline glue will be zero, and the natural height is $1+1-3+2=1\pt$ (because the depth of\/ |\box2| isn't included in the natural height); so the glue will ultimately become |\vskip-1pt| when it's set. Thus, |\box3| is $3\pt$ high, $2\pt$ deep, $4\pt$ wide. Its reference point coincides with that of\/ |\box2|; to get to the reference point of\/ |\box1| you go up $2\pt$ and right $3\pt$. \ddangerexercise Under the assumptions of the previous exercise, but with |\baselineskip=9pt minus3fil|, describe |\box4| after \begintt \setbox4=\vbox to4pt{\vss\box1\moveleft4pt\box2\vss} \endtt \answer The interline glue will be $6\pt$ minus $3\,{\rm fil}$; the final depth will be zero, since |\box2| is followed by glue; the natural height is $12\pt$; and the shrinkability is $5\,{\rm fil}$. So |\box4| will be $4\pt$ high, $0\pt$ deep, $1\pt$ wide, and it will contain five items: |\vskip|\penalty0\hbox{|-1.6pt|}, |\box1|, |\vskip1.2pt|, |\moveleft4pt\box2|, |\vskip-1.6pt|. Starting at the reference point of |\box4|, you get to the reference point of\/ |\box1| by going up $4.6\pt$, or to the reference point of\/ |\box2| by going up $.4\pt$ and left $4\pt$. \ (For example, you go up $4\pt$ to get to the upper left corner of |\box4|; then down $-1.6\pt$, i.e., up $1.6\pt$, to get to the upper left corner of\/ |\box1|; then down $1\pt$ to reach its reference point. This problem is clearly academic, since it's rather ridiculous to include infinite shrinkability in the baselineskip.) \ddangerexercise Solve the previous problem but with |\boxmaxdepth=-4pt|. \answer Now |\box4| will be $4\pt$ high, $-4\pt$ deep, $1\pt$ wide, and it will contain |\vskip|\penalty0\hbox{|-2.4pt|}, |\box1|, |\vskip-1.2pt|, |\moveleft4pt\box2|, |\vskip-2.4pt|. From the baseline of\/ |\box4|, go up exactly $5.4\pt$ to reach the baseline of\/ |\box1|, or exactly $3.6\pt$ to reach the baseline of\/ |\box2|. \danger We have observed that |\vbox| combines a bunch of boxes into a larger box that has the same baseline as the bottom box inside. \TeX\ has another operation called ^|\vtop|, which gives you a box like |\vbox| but with the same baseline as the top box inside. For example, \begintt \hbox{Here are \vtop{\hbox{two lines}\hbox{of text.}}} \endtt produces \begindisplay Here are \vtop{\hbox{two lines}\hbox{of text.}} \enddisplay \ddanger You can say `|\vtop to|\' and `|\vtop spread|\' just as with |\vbox|, but you should realize what such a construction means. \TeX\ implements |\vtop| as follows: \ (1)~First a~vertical box is formed as if\/ |\vtop| had been |\vbox|, using all of the rules for |\vbox| as given above. \ (2)~The final height~$x$ is defined to be zero unless the very first item inside the new vbox is a box; in the latter case, $x$ is the height of that box. \ (3)~Let $h$ and~$d$ be the height and depth of the vbox in step~(1). \TeX\ completes the |\vtop| by moving the reference point up or down, if necessary, so that the box has height~$x$ and depth $h+d-x$. \ddangerexercise Describe the empty boxes that you get from `|\vbox to|\|{}|' and `|\vtop to|\|{}|'. What are their heights, depths, and widths? \answer |\vbox to| $x$|{}| produces height $x$; |\vtop to| $x$|{}| produces depth $x$; the other dimensions are zero. \ (This holds even when $x$ is negative.) \ddangerexercise Define a macro |\nullbox#1#2#3| that produces a box whose height, depth, and width are given by the three parameters. The box should contain nothing that will show up in print. \answer There are several possibilities: \begintt \def\nullbox#1#2#3{\vbox to#1{\vss\hrule height-#2depth#2width#3}} \endtt works because the rule will be of zero thickness. Less tricky is \begintt \def\nullbox#1#2#3{\vbox to#1{\vss\vtop to#2{\vss\hbox to#3{}}}} \endtt Both of these are valid with negative height and/or depth, but they do not produce negative width. If the width might be negative, but not the height or depth, you can use, e.g., |\def\nullbox#1#2#3{\hbox to#3{\hss\raise#1\null\lower#2\null}}|. It's impossible for |\hbox| to construct a box whose height or depth is negative; it's impossible for |\vbox| or |\vtop| to construct a box whose width is negative.\par However, there's actually a trivial solution to the general problem, based on features that will be discussed later: \begintt \def\nullbox#1#2#3{\setbox0=\null \ht0=#1 \dp0=#2 \wd0=#3 \box0 } \endtt \danger The |\vbox| operation tends to produce boxes with large height and small depth, while |\vtop| tends to produce small height and large depth. If you're trying to make a vertical list out of big vboxes, however, you may not be satisfied with either |\vbox| or |\vtop|; you might well wish that a box had two reference points simultaneously, one for the top and one for the bottom. If such a dual-reference-point scheme were in use, one could define interline glue based on the distance between the lower reference point of one box and the upper reference point of its successor in a vertical list. But alas, \TeX\ gives you only one reference point per box. \danger There's a way out of this dilemma, using an important idea called a ``^{strut}.'' Plain \TeX\ defines ^|\strut| to be an invisible box of width zero that extends just enough above and below the baseline so that you would need no interline glue at all if every line contained a strut. \ (Baselines are $12\pt$ apart in plain \TeX; it turns out that |\strut| is a vertical rule, $8.5\pt$ high and $3.5\pt$ deep and $0\pt$ wide.) \ If you contrive to put a strut on the top line and another on the bottom line, inside your large vboxes, then it's possible to obtain the correct spacing in a larger assembly by simply letting the boxes butt together. For example, the ^|\footnote| macro in Appendix~B puts struts at the beginning and end of every footnote, so that the spacing will be right when several footnotes occur together at the bottom of some page. ^^{fitting boxes together} \danger If you understand boxes and glue, you're ready to learn the ^|\rlap| and ^|\llap| macros of plain \TeX; these names are abbreviations for ``right ^{overlap}'' and ``left overlap.'' Saying `|\rlap{|\|}|' is like typesetting \ and then backing up as if you hadn't typeset anything. More precisely, `|\rlap{|\|}|' creates a box of width zero, with `\' appearing just at the right of that box (but not taking up any space). The |\llap| macro is similar, but it does the ^{backspacing} first; in other words, `|\llap{|\|}|' creates a box of width zero, with `\' extending just to the left of that box. Using typewriter type, for example, you can typeset `{\tt\rlap/=}' by saying either `|\rlap/=|' or `|/\llap=|'. It's possible to put text into the left margin using |\llap|, or into the right margin using |\rlap|, because \TeX\ does not insist that the contents of a box must be strictly confined within that box's boundaries. ^^{marginal notes} \danger The interesting thing about |\rlap| and |\llap| is that they can be done so simply with infinite glue. One way to define |\rlap| would be \begintt \def\rlap#1{{\setbox0=\hbox{#1}\copy0\kern-\wd0}} \endtt but there's no need to do such a lengthy computation. The actual definition in Appendix~B is much more elegant, namely, \begintt \def\rlap#1{\hbox to 0pt{#1\hss}} \endtt and it's worth pondering why this works. Suppose, for example, that you're doing |\rlap{g}| where the letter `g' is $5\pt$ wide. Since |\rlap| makes an hbox of width $0\pt$, the glue represented by ^|\hss| must shrink by $5\pt$. Well, that glue has $0\pt$ as its natural width, but it has infinite shrinkability, so it can easily shrink to $-5\pt$; and `|\hskip-5pt|' is exactly what |\rlap| wants in this case. \dangerexercise Guess the definition of\/ |\llap|, without peeking at Appendices A or~B. \answer |\def\llap#1{\hbox to 0pt{\hss#1}}| \dangerexercise (This is a sequel to exercise 12.\linexno, but it's trickier.) \ Describe the result of \begintt \line{\hfil A puzzle.\hfilneg} \endtt \answer You get `A' at the extreme left and `puzzle.\null' at the extreme right, because the space between words has the only stretchability that is finite; the infinite stretchability cancels out. \ (In this case, \TeX's rule about ^{infinite glue} differs from what you would get in the limit if the value of $1\,{\rm fil}$ were finite but getting larger and larger. The true limiting behavior would stretch the text `A~puzzle.\null' in the same way, but it would also move that text infinitely far away beyond the right edge of the page.) \endchapter There was things which he stretched, but mainly he told the truth. \author MARK ^{TWAIN}{^^{Clemens}}, {\sl Huckleberry Finn\/} (1884) % chap1 \bigskip Every shape exists only because of the space around it. % Jeder Formwert lebt nur dank seiner Umgebung. $\ldots$ Hence there is a `right' position for every shape in every situation. % ... Daraus geht hervor, ein bestimmter Formvert in einer bestimmten Situation % einen bestimmten Platz verlangt, der ihm gem\"a\ss\ ist. If we succeed in finding that position, we have done our job. % Gelingt es uns, diese Pl\"atze zu finden, % so nennen wir die Arbeit vollendet. \author JAN ^{TSCHICHOLD}, {\sl Typographische Gestaltung\/} (1935) % p64 % I've borrowed from the free translation in {\sl Asymmetric Typography} \eject \beginchapter Chapter 13. Modes Just as people get into different moods, \TeX\ gets into different ``modes.'' \ (Except that \TeX\ is more predictable than people.) \ There are six ^{modes}: \medskip \item\bull^{Vertical mode}. [Building the main vertical list, from which the pages of output are derived.] \smallskip\item\bull ^{Internal vertical mode}. [Building a vertical list for a vbox.] \smallskip\item\bull ^{Horizontal mode}. [Building a horizontal list for a paragraph.] \smallskip\item\bull ^{Restricted horizontal mode}. [Building a horizontal list for an hbox.] \smallskip\item\bull ^{Math mode}. [Building a mathematical formula to be placed in a horizontal list.] \smallskip\item\bull ^{Display math mode}. [Building a mathematical formula to be placed on a line by itself, temporarily interrupting the current paragraph.] \medskip\noindent In simple situations, you don't need to be aware of what mode \TeX\ is in, because the computer just does the right thing. But when you get an error message that says `\thinspace|!|~|You| |can't| |do| |such-and-such| |in| |restricted| |horizontal| |mode|\thinspace', a knowledge of modes helps to explain why \TeX\ thinks you goofed. Basically \TeX\ is in one of the vertical modes when it is preparing a list of boxes and glue that will be placed vertically above and below one another on the page; it's in one of the horizontal modes when it is preparing a list of boxes and glue that will be strung out horizontally next to each other with baselines aligned; and it's in one of the math modes when it is reading a formula. A play-by-play account of a typical \TeX\ job should make the mode idea clear: At the beginning, \TeX\ is in vertical mode, ready to construct pages. If you specify glue or a box when \TeX\ is in vertical mode, the glue or the box gets placed on the current page below what has already been specified. For example, the ^|\vskip| instructions in the sample run we discussed in Chapter~6 contributed vertical glue to the page; and the ^|\hrule| instructions contributed horizontal rules at the top and bottom of the story. The ^|\centerline| commands also produced boxes that were included in the main vertical list; but those boxes required a bit more work than the rule boxes: \TeX\ was in vertical mode when it encountered `|\centerline{\bf A SHORT STORY}|', and it went temporarily into restricted horizontal mode while processing the words `|A SHORT STORY|'; then the digestive process returned to vertical mode, after setting the glue in the~|\centerline|~box. Continuing with the example of Chapter 6, \TeX\ switched into horizontal mode as soon as it read the `|O|' of `|Once upon a time|'. Horizontal mode is the mode for making ^{paragraphs}. The entire paragraph (lines 7 to~11 of the |story| file) was input in horizontal mode; then the text was divided into output lines of the appropriate width, those lines were put in boxes and appended to the page (with appropriate interline glue between them), and \TeX\ was back in vertical mode. The `|M|' on line~12 started up horizontal mode again. When \TeX\ is in vertical mode or internal vertical mode, the first token of a new paragraph changes the mode to horizontal for the duration of a paragraph. In other words, things that do not have a vertical orientation cause the mode to switch automatically from vertical to horizontal. This occurs when you type any character, or ^|\char| or ^|\accent| or ^|\hskip| or |\|\] ^^{control space} or ^|\vrule| or math shift (|$|); \TeX\ inserts the current paragraph ^{indentation} and rereads the horizontal token as if it had occurred in horizontal mode. \danger You can also tell \TeX\ explicitly to go into horizontal mode, instead of relying on such implicit mode-switching, by saying `^|\indent|' or `^|\noindent|'. For example, if line~7 of the |story| file in Chapter~6 had begun \begintt \indent Once upon a time, ... \endtt the same output would have been obtained, because `|\indent|' would have instructed \TeX\ to begin the paragraph. And if that line had begun with \begintt \noindent Once upon a time, ... \endtt the first paragraph of the story would not have been indented. The |\noindent| command simply tells \TeX\ to enter horizontal mode if the current mode is vertical or internal vertical; |\indent| is similar, but it also creates an empty box whose width is the current value of ^|\parindent|, and it puts this empty box into the current horizontal list. Plain \TeX\ sets |\parindent=20pt|. If you say |\indent\indent|, you get double indentation; if you say |\noindent\noindent|, the second |\noindent| does nothing. \dangerexercise If you say `^|\hbox||{...}|' in horizontal mode, \TeX\ will construct the specified box and it will contribute the result to the current paragraph. Similarly, if you say `|\hbox{...}|' in vertical mode, \TeX\ will construct a box and contribute it to the current page. What can you do if you want to begin a paragraph with an |\hbox|? \answer Simply saying |\hbox{...}| won't work, since that box will just continue the previous vertical list without switching modes. You need to start the paragraph explicitly, and the straightforward way to do that is to say |\indent\hbox{...}|. But suppose you want to define a macro that expands to an hbox, where this macro is to be used in the midst of a paragraph as well as at the beginning; then you don't want to force users to type |\indent| before calling your macro at the beginning of a paragraph, nor do you want to say |\indent| in the macro itself (since that might insert unwanted indentations). One solution to this more general problem is to say `|\|\]^|\unskip||\hbox{...}|', since |\|\] makes the mode horizontal while |\unskip| removes the unwanted space. Plain \TeX\ provides a ^|\leavevmode| macro, which solves this problem in what is probably the most efficient way: |\leavevmode| is an abbreviation for `|\unhbox\voidbox|', where |\voidbox| is a permanently empty box register. When handling simple manuscripts, \TeX\ spends almost all of its time in horizontal mode (making paragraphs), with brief excursions into vertical mode (between paragraphs). A paragraph is completed when you type ^|\par| or when your manuscript has a blank line, since a blank line is converted to |\par| by the reading rules of Chapter~8. A paragraph also ends when you type certain things that are incompatible with horizontal mode. For example, the command `|\vskip 1in|' on line~16 of Chapter~6's |story| file was enough to terminate the paragraph about `|...beautiful documents.|'; no |\par| was necessary, since |\vskip| introduced vertical glue that couldn't belong to the~paragraph. If a begin-math token (|$|) appears in horizontal mode, \TeX\ plunges into math mode and processes the formula up until the closing `|$|', then appends the text of this formula to the current paragraph and returns to horizontal mode. Thus, in the ``I wonder why?''\ example of Chapter~12, \TeX\ went into math mode temporarily while processing |\ldots|, treating the dots as a formula. However, if two consecutive begin-math tokens appear in a paragraph (|$$|), \TeX\ interrupts the paragraph where it is, contributes the paragraph-so-far to the enclosing vertical list, then processes a math formula in display math mode, then contributes this formula to the enclosing list, then returns to horizontal mode for more of the paragraph. \ (The formula to be displayed should end with `|$$|'.) \ For example, suppose you type \begintt the number $$\pi \approx 3.1415926536$$ is important. \endtt \TeX\ goes into display math mode between the |$$|'s, and the output you get states that the number $$\pi \approx 3.1415926536$$ is important. ^^|\pi| \smallskip \TeX\ ignores blank spaces and blank lines (or ^|\par| commands) when it's in vertical or internal vertical mode, so you need not worry that such things might change the mode or affect a printed document. A ^{control space} (|\|\]) will, however, be regarded as the beginning of a paragraph; the paragraph will start with a blank space after the indentation. \smallskip At the end of a \TeX\ manuscript it's usually best to finish everything off by typing `^|\bye|', which is plain \TeX's abbreviation for `|\vfill\eject\end|'. The `|\vfill|' gets \TeX\ into vertical mode and inserts enough space to fill up the last page; `|\eject|' outputs that last page; and `|\end|' sends the computer into its ^{endgame} routine. \danger \TeX\ gets into internal vertical mode when you ask it to construct something from a vertical list of boxes (using |\vbox| or |\vtop| or |\vcenter| or |\valign| or |\vadjust| or |\insert|). It gets into restricted horizontal mode when you ask it to construct something from a horizontal list of boxes (using |\hbox| or |\halign|). Box construction is discussed in Chapters 12 and~21. We will see later that there is very little difference between internal vertical mode and ordinary vertical mode, and very little difference between restricted horizontal mode and ordinary horizontal mode; but they aren't quite identical, because they have different goals. \danger Whenever \TeX\ looks at a token of input to decide what should be done next, the current mode has a potential influence on what that token means. For example, ^|\kern| specifies vertical spacing in vertical mode, but it specifies horizontal spacing in horizontal mode; a math shift character like `|$|' causes entry to math mode from horizontal mode, but it causes exit from math mode when it occurs in math mode; two consecutive math shifts (|$$|) appearing in horizontal mode will initiate display math mode, but in restricted horizontal mode they simply denote an empty math formula. \TeX\ uses the fact that some operations are inappropriate in certain modes to help you recover from errors that might have crept into your manuscript. Chapters 24 to~26 explain exactly what happens to every possible token in every possible mode. \danger \TeX\ often interrupts its work in one mode to do some task in another mode, after which the original mode is resumed again. For example, you can say `|\hbox{|' in any mode; when \TeX\ digests this, it suspends whatever else it was doing and enters restricted horizontal mode. The matching `|}|' will eventually cause the hbox to be completed, whereupon the postponed task will be taken up anew. In this sense \TeX\ can be in many modes simultaneously, but only the innermost mode influences the calculations at any time; the other modes have been pushed out of \TeX's consciousness. \goodbreak \danger One way to become familiar with \TeX's modes is to consider the following curious test file called |modes.tex|, which exercises all the modes at once: $$\halign{\hbox to\parindent{\hfil\sevenrm#\ \ }&#\hfil\cr 1&|\tracingcommands=1|\cr\noalign{^^|\tracingcommands|} 2&|\hbox{|\cr 3&|$|\cr 4&|\vbox{|\cr 5&|\noindent$$|\cr 6&|x\showlists|\cr 7&|$$}$}\bye|\cr}$$ The first line of ^|modes.tex| tells \TeX\ to log every command it receives; \TeX\ will produce diagnostic data whenever |\tracingcommands| is positive. Indeed, if you run \TeX\ on |modes.tex| you will get a |modes.log| file that includes the following information: \begintt {vertical mode: \hbox} {restricted horizontal mode: blank space } {math shift character $} {math mode: blank space } {\vbox} {internal vertical mode: blank space } {\noindent} {horizontal mode: math shift character $} {display math mode: blank space } {the letter x} \endtt The meaning is that \TeX\ first saw an |\hbox| token in vertical mode; this caused it to go ahead and read the `|{|' behind the scenes. Then \TeX\ entered restricted horizontal mode, and saw the blank space token that resulted from the end of line~2 in the file. Then it saw a math shift character token (still in restricted horizontal mode), which caused a shift to math mode; another blank space came through. Then |\vbox| inaugurated internal vertical mode, and |\noindent| instituted horizontal mode within that; two subsequent |$| signs led to display math mode. \ (Only the first |$| was shown by |\tracingcommands|, because that one caused \TeX\ to look ahead for another.) \danger The next thing in |modes.log| after the output above is `|{|^|\showlists||}|'. This is another handy diagnostic command that you can use to find out things that \TeX\ ordinarily keeps to itself; it causes \TeX\ to display the lists that are being worked on, in the current mode and in all enclosing modes where the work has been suspended: \begintt ### display math mode entered at line 5 \mathord .\fam1 x ### internal vertical mode entered at line 4 prevdepth ignored ### math mode entered at line 3 ### restricted horizontal mode entered at line 2 \glue 3.33333 plus 1.66666 minus 1.11111 spacefactor 1000 ### vertical mode entered at line 0 prevdepth ignored \endtt In this case the lists represent five levels of activity, all present at the end of line~6 of |modes.tex|. The current mode is shown first, namely, display math mode, which began on line~5. The current math list contains one ``^{mathord}'' object, consisting of the letter~|x| in family~1. \ (Have patience and you will understand what that means, when you learn about \TeX's math formulas.) \ Outside of display math mode comes internal vertical mode, to which \TeX\ will return when the paragraph containing the displayed formula is complete. The vertical list ^^|prevdepth ignored| on that level is empty; `|prevdepth ignored|' means that |\prevdepth| has a value $\le-1000\pt$, so that the next interline glue will be omitted (cf.~Chapter~12). The math mode outside of this internal vertical mode has an empty list, likewise, but the restricted horizontal mode enclosing the math mode contains some glue. Finally, we see the main vertical mode that encloses everything; this mode was `|entered at line 0|', i.e., before the file |modes.tex| was input; nothing has been contributed so far to the vertical list on this outermost level. \dangerexercise Why is there glue in one of these lists but not in the others? \answer The output of\/ |\tracingcommands| shows that four blank space tokens were digested; these originated at the ends of lines 2,~3, 4, and~5. Only the first had any effect, since blank spaces are ignored in math formulas and in vertical modes. \dangerexercise After this output of\/ |\showlists|, the |modes.log| file contains further output from |\tracingcommands|. In fact, the next two lines of that file are \begintt {math shift character $} {horizontal mode: end-group character }} \endtt because the `|$$|' on line 7 finishes the displayed formula, and this resumes horizontal mode for the paragraph that was interrupted. What do you think are the next three lines of |modes.log|\thinspace? \answer The |end-group character| finishes the paragraph and the |\vbox|, and |\bye| stands for `|\par\vfill...|', so the next three commands are \begintt {math mode: math shift character $} {restricted horizontal mode: end-group character }} {vertical mode: \par} \endtt \dangerexercise Suppose \TeX\ has generated a document without ever leaving vertical mode. What can you say about that document? \answer It contains only mixtures of vertical glue and horizontal rules whose reference points appear at the left of the page; there's no text. \ddangerexercise Some of \TeX's modes cannot immediately enclose other modes; for example, display math mode is never directly enclosed by horizontal mode, even though displays occur within paragraphs, because an interrupted paragraph-so-far of horizontal mode is always completed and removed from \TeX's memory before the processing of a displayed formula begins. Give a complete characterization of all pairs of consecutive modes that can occur in the output of\/ |\showlists|. \answer Vertical mode can occur only as the outermost mode; horizontal mode and display math mode can occur only when immediately enclosed by vertical or internal vertical mode; ordinary math mode cannot be immediately enclosed by vertical or internal vertical mode; all other cases are possible. \endchapter Every mode of life has its conveniences. \author SAMUEL ^{JOHNSON}, {\sl The Idler\/} (1758) \bigskip [Hindu musicians] have eighty-four modes, of which thirty-six are in general use, and each of which, it appears, has a peculiar expression, and the power of moving some particular sentiment or affection. \author MOUNTSTUART ^{ELPHINSTONE}, {\sl History of India\/} (1841) % III.vii.I.297 \eject \beginchapter Chapter 14. How \TeX\ Breaks\\Paragraphs into Lines One of a typesetting system's chief duties is to take a long sequence of words and to break it up into individual lines of the appropriate size. For example, every paragraph of this manual has been broken into lines that are 29~picas wide, but the author didn't have to worry about such details when he composed the manuscript. \TeX\ chooses breakpoints in an interesting way that considers each paragraph in its entirety; the closing words of a paragraph can actually influence the appearance of the first line. As a result, the spacing between words is as uniform as possible, and the computer is able to reduce the number of times that words must be hyphenated or formulas must be split between lines. ^^{H\&J, see hyphenation, line breaking, setting glue} ^^{justification, see setting glue, line breaking} ^^{quad left, see flush left} ^^{quad right, see flush right} ^^{quad middle, see :break} The experiments of Chapter 6 have already illustrated the general ideas: We discussed the notion of ``badness,'' and we ran into ``overfull'' and ``underfull'' boxes in difficult situations. We also observed that different settings of \TeX's ^|\tolerance| parameter will produce different effects; a higher tolerance means that wider spaces are acceptable. \TeX\ will find the absolutely best way to typeset any given paragraph, according to its ideas of minimum badness. But such ``badness'' doesn't account for everything, and if you rely entirely on an automatic scheme you will occasionally encounter line breaks that are not really the best on psychological grounds; this is inevitable, because computers don't understand things the way people~do (at least not yet). Therefore you'll sometimes want to tell the machine that certain places are not good breakpoints. Conversely, you will sometimes want to force a break at a particular spot. \TeX\ provides a convenient way to avoid psychologically bad breaks, so that you will be able to obtain results of the finest quality by simply giving a few hints to the machine. ``^{Ties}''---denoted by `|~|' in plain \TeX---are the key to successful line breaking. ^^{auxiliary space, see tie} ^^{tilde} Once you learn how to insert them, you will have graduated from the ranks of ordinary \TeX nical typists to the select group of Distinguished \TeX nicians. And it's really not difficult to train yourself to insert occasional ties, almost without thinking, as you type a manuscript. ^^{line breaks, avoiding} ^^{breaks, avoiding bad} When you type |~| it's the same as typing a space, except that \TeX\ won't break a line at this space. Furthermore, you shouldn't leave any blanks next to the |~|, since they will count as additional spaces. If you put |~| at the very end of a line in your input file, you'll get a wider space than you want, because the \ that follows the |~| produces an extra space. We have already observed in Chapter~12 that it's generally a good idea to type |~| after an abbreviation that does not come at the end of a sentence. Ties also belong in several other places: \smallskip \item\bull In references to named parts of a document: $$\halign{#\hfil&\hskip 80pt#\hfil\cr |Chapter~12|&|Theorem~1.2|\cr |Appendix~A|&|Table~\hbox{B-8}|\cr |Figure~3|&|Lemmas 5 and~6|\cr}$$ \noindent(No |~| appears after `|Lemmas|' in the final example, since there's no harm in having `5~and~6' at the beginning of a line. The use of\/ |\hbox| is explained below.) \smallbreak \item\bull Between a person's forenames and between multiple surnames: $$\halign{#\hfil&\hskip 40pt#\hfil\cr |Donald~E. Knuth|&|Luis~I. Trabb~Pardo|\cr |Bartel~Leendert van~der~Waerden|&|Charles~XII|\cr}$$ ^^{Knuth} ^^{Trabb Pardo} ^^{van der Waerden} ^^{Charles XII} \noindent Note that it is sometimes better to hyphenate a name than to break it between words; e.g., `Don-' and `ald~E.~Knuth' is more tolerable than `Donald' and `E.~Knuth'. The previous rule can be regarded as a special case of this one, since we may think of `Chapter~12' as a compound name; another example is `|register~X|'. Sometimes a name is so long that we dare not tie it all together, lest there be no way to break the line: ^^{Vall\'ee Poussin} \begintt Charles Louis Xavier~Joseph de~la Vall\'ee~Poussin. \endtt \item\bull Between math symbols in apposition with nouns: $$\halign{#\hfil\cr |dimension~$d$ width~$w$ function~$f(x)$|\cr |string~$s$ of length~$l$|\cr}$$ \noindent However, the last example should be compared with \begintt string~$s$ of length $l$~or more. \endtt \item\bull Between symbols in series: $$\halign{#\hfil\cr |1,~2, or~3|\cr |$a$,~$b$, and~$c$.|\cr |1,~2, \dots,~$n$.|\cr}$$ \item\bull When a symbol is a tightly bound object of a preposition: $$\halign{#\hfil\cr |of~$x$|\cr |from 0 to~1|\cr |increase $z$ by~1|\cr |in common with~$m$.|\cr}$$ \noindent The rule does not, however, apply to compound objects: \begintt of $u$~and~$v$. \endtt \item\bull When mathematical phrases are rendered in words: $$\halign{#\hfil&\hskip20pt#\hfil&\hskip20pt#\hfil\cr |equals~$n$|&|less than~$\epsilon$|&|(given~$X$)|\cr |mod~2|&|modulo~$p^e$|&|for all large~$n$|\cr}$$ \noindent Compare `|is~15|' with `|is 15~times the height|'. \medbreak \item\bull When cases are being enumerated within a paragraph: ^^{enumerated cases within a paragraph} $$\halign{#\hfil\cr |(b)~Show that $f(x)$ is (1)~continuous; (2)~bounded.|\cr}$$ \noindent It would be nice to boil all of these rules down to one or two simple principles, and~it would be even nicer if the rules could be automated so that keyboarding could be done without them; but subtle semantic considerations seem to be involved. Therefore it's best to use your own judgment with respect to ties. The computer needs your help. A tie keeps \TeX\ from breaking at a space, but sometimes you want to prevent the machine from breaking at a ^{hyphen} or a ^{dash}. This can be done by using ^|\hbox|, because \TeX\ will not split up the contents of a box; boxes are indecomposable units, once they have been constructed. We have already illustrated this principle in the `|Table~\hbox{B-8}|' example considered earlier. Another example occurs when you are typing the page numbers in a ^{bibliographic reference}: It doesn't look good to put \hbox{`22.'} on a line by itself, so you can type `|\hbox{13--22}.|' to prohibit breaking `\hbox{13--22}.' On the other hand, \TeX\ doesn't often choose line breaks at hyphens, so you needn't bother to insert |\hbox| commands unless you need to correct a bad break that \TeX\ has already made on a previous run. \exercise Here are some phrases culled from previous chapters of this manual. How do you think the author typed them? \begindisplay (cf.~Chapter~12).\cr Chapters 12 and~21.\cr line~16 of Chapter~6's {\tt story}\cr lines 7 to~11\cr lines 2,~3, 4, and~5.\cr (2)~a big black bar\cr All 256~characters are initially of category~12,\cr letter~{\tt x} in family~1.\cr the factor~$f$, where $n$~is 1000~times~$f$.\cr \enddisplay \answer|(cf.~Chapter~12).|\parbreak |Chapters 12 and~21.|\parbreak |line~16 of Chapter~6's {\tt story}|\parbreak |lines 7 to~11|\parbreak |lines 2,~3, 4, and~5.|\parbreak |(2)~a big black bar|\parbreak |All 256~characters are initially of category~12,|\parbreak |letter~{\tt x} in family~1.|\parbreak |the factor~$f$, where $n$~is 1000~times~$f$.| \exercise How would you type the phrase `for all $n$ greater than $n_0$'\thinspace? \answer `|for all $n$~greater than~$n_0$|' avoids distracting breaks. \exercise And how would you type `exercise 4.3.2--15'\thinspace? \answer `|exercise \hbox{4.3.2--15}|' guarantees that there is no break after the ^{en-dash}. But this precaution is rarely necessary, so `|exercise 4.3.2--15|' is an acceptable answer. No |~| is needed; `\hbox{4.3.2--15}' is so long that it causes no offense at the beginning of a line. \exercise Why is it better to type `|Chapter~12|' than to type `|\hbox{Chapter 12}|'\thinspace? \answer The space you get from |~| will stretch or shrink with the other spaces in the same line, but the space inside an hbox has a fixed width since that glue has already been set once and for all. Furthermore the first alternative permits the word Chap-\break ter to be ^{hyphenate}d. \dangerexercise \TeX\ will sometimes break a math formula after an equals sign. How can you stop the computer from breaking the formula `$x=0$'\thinspace? \answer `|\hbox{$x=0$}|' is unbreakable, and we will see later that `|${x=0}$|' cannot be broken. Both of these solutions set the glue surrounding the equals sign to some fixed value, but such glue normally wants to stretch; furthermore, the |\hbox| solution might include undesirable blank space at the beginning or end of a line, if\/ ^|\mathsurround| is nonzero. A third solution `|$x=\nobreak0$|' avoids both defects. \ddangerexercise Explain how you could instruct \TeX\ not to make any breaks after explicit hyphens and dashes. \ (This is useful in lengthy ^{bibliographies}.) \answer |\exhyphenpenalty=10000| prohibits all such breaks, according to the rules found later in this chapter. Similarly, |\hyphenpenalty=10000| prevents breaks after implicit (discretionary) hyphens. Sometimes you want to permit a line break after a `/' just as if it were a hyphen. For this purpose plain \TeX\ allows you to say `^|\slash|'; for example, `|input\slash output|' produces `input\slash output' with an optional break. {\hbadness=10000 If you want to force \TeX\ to break between lines at a certain point in ^^{line breaks, forcing} ^^{breaks, forcing good} the middle of a paragraph, just say `^|\break|'. However, that might cause the line to be really ^^{underfull}spaced out.\break If you want \TeX\ to fill up the right-hand part of a line with blank space just before a forced line break,\hfil\break without indenting the next line, say `|\hfil\break|'.\par} \danger You may have several consecutive lines of input for which you want the output to appear line-for-line in the same way. One solution is to type `|\par|' at the end of each input line; but that's somewhat of a nuisance, so plain \TeX\ provides the abbreviation `^|\obeylines|', which causes each end-of-line in the input to be like |\par|. After you say |\obeylines| you will get one line of output per line of input, unless an input line ends with `|%|' or unless it is so long that it must be broken. For example, you probably want to use |\obeylines| if you are typesetting a ^{poem}. Be sure to enclose |\obeylines| in a group, unless you want this ``poetry mode'' to continue to the end of your document. \begintt {\obeylines\smallskip Roses are red, \quad Violets are blue; Rhymes can be typeset \quad With boxes and glue. \smallskip} \endtt \dangerexercise Explain the uses of\/ ^|\quad| in this poem. What would have happened if `|\quad|' had been replaced by `^|\indent|' in both places? \answer The second and fourth lines are indented by an additional ``quad'' of space, i.e., by one extra em in the current type style. \ (The control sequence |\quad| does an ^|\hskip|; when \TeX\ is in vertical mode, |\hskip| begins a new paragraph and puts glue after the indentation.) \ If\/ |\indent| had been used instead, those lines wouldn't have been indented any more than the first and third, because |\indent| is implicit at the beginning of every paragraph. Double indentation on the second and fourth lines could have been achieved by `|\indent\indent|'. Roughly speaking, \TeX\ breaks paragraphs into lines in the following way: Breakpoints are inserted between words or after hyphens so as to produce lines whose badnesses do not exceed the current ^|\tolerance|. If there's no way to insert such breakpoints, an ^{overfull box} is set. Otherwise the breakpoints are chosen so that the paragraph is mathematically optimal, i.e., best possible, in the sense that it has no more ``^{demerits}'' than you could obtain by any other sequence of breakpoints. Demerits are based on the badnesses of individual lines and on the existence of such things as consecutive lines that end with hyphens, or tight lines that occur next to loose ones. \danger But the informal description of line breaking in the previous paragraph is an oversimplification of what really happens. The remainder of this chapter explains the details precisely, for people who want to apply \TeX\ in nonstandard ways. \TeX's line-breaking algorithm has proved to be general enough to handle a surprising variety of different applications; this, in fact, is probably the most interesting aspect of the whole \TeX\ system. However, every paragraph from now on until the end of the chapter is prefaced by at least one dangerous bend sign, so you may want to learn the following material in easy stages instead of all at once. \ninepoint \danger Before the lines have been broken, a paragraph inside of \TeX\ is actually a {\sl ^{horizontal list}}, i.e., a sequence of items that \TeX\ has gathered while in horizontal mode. We have been saying informally that a horizontal list consists of boxes and glue; the truth is that boxes and glue aren't the whole story. Each item in a horizontal list is one of the following types of things:\enddanger \smallskip \item\bull a box (a character or ligature or rule or hbox or vbox); \item\bull a ^{discretionary break} (to be explained momentarily); ^^{break, discretionary} \item\bull a ``^{whatsit}'' (something special to be explained later); \item\bull vertical material (from ^|\mark| or ^|\vadjust| or ^|\insert|); \item\bull a glob of ^{glue} (or ^|\leaders|, as we will see later); \item\bull a ^{kern} (something like glue that doesn't stretch or shrink); \item\bull a ^{penalty} (representing the undesirability of breaking here); \item\bull ``^{math-on}'' (beginning a formula) or ``^{math-off}'' (ending a formula). \smallskip\noindent The last four types (glue, kern, penalty, and math items) are called {\sl ^{discardable}}, since they may change or disappear at a line break; the first four types are called non-discardable, since they always remain intact. Many of the things that can appear in horizontal lists have not been touched on yet in this manual, but it isn't necessary to understand them in order to understand line breaking. Sooner or later you'll learn how each of the gismos listed above can infiltrate a horizontal list; and if you want to get a thorough understanding of \TeX's internal processes, you can always use ^|\showlists| with various features of the language, in order to see exactly what \TeX\ is doing. \danger A discretionary break consists of three sequences of characters called the {\sl pre-break}, {\sl post-break}, and {\sl no-break\/} texts. The idea is that if a line break occurs here, the ^{pre-break text} will appear at the end of the current line and the ^{post-break text} will occur at the beginning of the next line; but if no break occurs, the ^{no-break text} will appear in the current line. Users can specify ^^|\discretionary| discretionary breaks in complete generality by writing \begindisplay |\discretionary{|\|}{|\|}{|\|}| \enddisplay where the three texts consist entirely of characters, boxes, and kerns. For example, \TeX\ can hyphenate the word `difficult' between the f's, even though this requires breaking the `ffi' ligature into `f-' followed by an `fi' ligature, if the horizontal list contains \begintt di\discretionary{f-}{fi}{ffi}cult. \endtt Fortunately you need not type such a mess yourself; \TeX's hyphenation algorithm works behind the scenes, taking ^{ligatures} apart and putting them into discretionary breaks when necessary. \danger The most common case of a discretionary break is a simple discretionary hyphen \begintt \discretionary{-}{}{} \endtt for which \TeX\ accepts the abbreviation `^|\-|'. The next most common case is \begintt \discretionary{}{}{} \endtt (an ``^{empty discretionary}''), which \TeX\ automatically inserts after `|-|' and after every ligature that ends with `|-|'. In the case of plain \TeX, empty discretionaries are therefore inserted after ^{hyphens} and ^{dashes}. \ (Each font has an associated ^|\hyphenchar|, which we can assume for simplicity is equal to `|-|'.) \danger When \TeX\ ^{hyphenates} words, it simply inserts discretionary breaks into the horizontal list. For example, the words `|discretionary hyphens|' are transformed into the equivalent of \begintt dis\-cre\-tionary hy\-phens \endtt if hyphenation becomes necessary. But \TeX\ doesn't apply its hyphenation algorithm to any word that already contains a discretionary break; therefore you can use explicit discretionaries to override \TeX's automatic method, in an emergency. \dangerexercise Before 1998, some ^{German} words changed their spelling when split between lines. For example, `backen' became `bak-ken' and `Bettuch' sometimes became `Bett-tuch'. How can you instruct \TeX\ to produce such effects? \answer |ba\ck/en| and |Be\ttt/uch|, where the macros |\ck/| and |\ttt/| are defined by \begintt \def\ck/{\discretionary{k-}{k}{ck}} \def\ttt/{tt\discretionary{-}{t}{}} \endtt The English word `eighteen' might deserve similar treatment. \TeX's hyphenation algorithm will not make such spelling changes automatically. \danger In order to save time, \TeX\ tries first to break a paragraph into lines without inserting any discretionary hyphens. This first pass will succeed if a sequence of breakpoints is found for which none of the resulting lines has a badness exceeding the current value of ^|\pretolerance|. If the first pass fails, the method of Appendix~H is used to hyphenate each word of the paragraph by inserting discretionary breaks into the horizontal list, and a second attempt is made using ^|\tolerance| instead of\/ |\pretolerance|. When the lines are fairly wide, as they are in this manual, experiments show that the first pass succeeds more than 90\% of the time, and that fewer than 2~words per paragraph need to be subjected to the hyphenation algorithm, on the average. But when the lines are very narrow the first pass usually fails rather quickly. Plain \TeX\ sets |\pretolerance=100| and |\tolerance=200| as the default values. If you make |\pretolerance=10000|, the first pass will essentially always succeed, so hyphenations will not be tried (and the spacing may be terrible); on the other hand if you make |\pretolerance=-1|, \TeX\ will omit the first pass and will try to hyphenate immediately. \danger Line breaks can occur only in certain places within a horizontal list. Roughly speaking, they occur between words and after hyphens, but in actuality they are permitted in the following five cases:\enddanger \smallskip \item{a)} at glue, provided that this glue is immediately preceded by a non-discardable item, and that it is not part of a math formula (i.e., not between math-on and math-off). A break ``at glue'' occurs at the left edge of the glue space. \smallskip \item{b)} at a kern, provided that this kern is immediately followed by glue, and that it is not part of a math formula. \smallskip \item{c)} at a math-off that is immediately followed by glue. \smallskip \item{d)} at a penalty (which might have been inserted automatically in a formula). \smallskip \item{e)} at a discretionary break. \smallskip\noindent Notice that if two globs of glue occur next to each other, the second one will never be selected as a breakpoint, since it is preceded by glue (which is discardable). \danger Each potential breakpoint has an associated ``penalty,'' which represents the ``aesthetic cost'' of breaking at that place. In cases (a), (b),~(c), the penalty is zero; in case~(d) an explicit penalty has been specified; and in case~(e) the penalty is the current value of ^|\hyphenpenalty| if the pre-break text is nonempty, or the current value of ^|\exhyphenpenalty| if the pre-break text is empty. Plain \TeX\ sets |\hyphenpenalty=50| and |\exhyphenpenalty=50|. \danger For example, if you say `^|\penalty| |100|' at some point in a paragraph, that position will be a legitimate place to break between lines, but a penalty of 100 will be charged. If you say `\hbox{|\penalty-100|}' you are telling \TeX\ that this is a rather good place to break, because a negative penalty is really a ``^{bonus}''; a line that ends with a bonus might even have ``merits'' (negative demerits). \danger Any penalty that is 10000 or more is considered to be so large ^^{infinite penalty} that \TeX\ will never break there. At the other extreme, any penalty that is $-10000$ or less is considered to be so small that \TeX\ will always break there. The ^|\nobreak| macro of plain \TeX\ is simply an abbreviation for `|\penalty10000|', because this prohibits a line break. A tie in plain \TeX\ is equivalent to `|\nobreak\|\]'; there will be no break at the glue represented by |\|\] in this case, because glue is never a legal breakpoint when it is preceded by a discardable item like a penalty. \dangerexercise Guess how the ^|\break| macro is defined in plain \TeX. \answer |\def\break{\penalty-10000 }| \dangerexercise What happens if you say |\nobreak\break| or |\break\nobreak|? \answer You get a forced break as if\/ |\nobreak| were not present, because |\break| cannot be cancelled by another penalty. In general if you have two penalties in a row, their combined effect is the same as a single penalty whose value is the minimum of the two original values, unless both of those values force breaks. \ (You get two breaks from |\break\break|; the second one creates an empty line.) \danger When a line break actually does occur, \TeX\ removes all discardable items that follow the break, until coming to something non-discardable, or until coming to another chosen breakpoint. For example, a sequence of glue and penalty items will vanish as a unit, if no boxes intervene, unless the optimum breakpoint sequence includes one or more of the penalties. Math-on and math-off items act essentially as kerns that contribute the spacing specified by ^|\mathsurround|; such spacing will disappear into the line break if a formula comes at the very end or the very beginning of a line, because of the way the rules have been formulated above. \ddanger The ^{badness} of a line is an integer that is approximately 100 times the cube of the ratio by which the glue inside the line must stretch or shrink to make an hbox of the required size. For example, if the line has a total shrinkability of 10 points, and if the glue is being compressed by a total of 9 points, the badness is computed to be~73 (since $100\times(9/10)^3=72.9$); similarly, a line that stretches by twice its total stretchability has a badness of 800. But if the badness obtained by this method turns out to be more than 10000, the value 10000 is used. \ $\bigl($See the discussion of ``^{glue set ratio}''~$r$ and ``^{glue set order}''~$i$ in Chapter~12; if $i\ne0$, there is infinite stretchability or shrinkability, so the badness is zero, otherwise the badness is approximately $\min(100r^3,10000)$.$\bigr)$ \ Overfull boxes are considered to be infinitely bad; they are avoided whenever possible. ^^{infinite badness} \ddanger A line whose badness is 13 or more has a glue set ratio exceeding 50\%. We call such a line {\sl ^{tight}\/} if its glue had to shrink, {\sl ^{loose}\/} if its glue had to stretch, and {\sl ^{very loose}\/} if it had to stretch so much that the badness is 100 or more. But if the badness is 12 or less we say that the line is {\sl ^{decent}}. Two adjacent lines are said to be {\sl {visually incompatible}\/} if their classifications are not adjacent, i.e., if a tight line is next to a loose or very loose line, or if a decent line is next to a very loose one. \ddanger \TeX\ rates each potential sequence of breakpoints by totalling up {\sl ^{demerits}\/} that are assessed to individual lines. The goal is to choose breakpoints that yield the fewest total demerits. Suppose that a line has badness~$b$, and suppose that the penalty~$p$ is associated with the breakpoint at the end of this line. As stated above, \TeX\ will not even consider such a line if $p\ge10000$, or if $b$~exceeds the current tolerance or pretolerance. Otherwise the demerits of such a line are defined by the formula \begindisplay $\displaystyle{d=\cases{ (l+b)^2+p^2,&if $0\le p<10000$;\cr (l+b)^2-p^2,&if $-10000 @@0|\cr |ferred to be called---was hap-pi-est when |\cr |@ via @@1 b=131 p=0 d=29881|\cr |@@2: line 2.0 t=32481 -> @@1|\cr |he|\cr |@ via @@1 b=25 p=0 d=1225|\cr |@@3: line 2.3 t=3825 -> @@1|\cr |was at work type-set-ting beau-ti-ful doc-|\cr |@\discretionary via @@2 b=1 p=50 d=12621|\cr |@\discretionary via @@3 b=291 p=50 d=103101|\cr |@@4: line 3.2- t=45102 -> @@2|\cr |u-|\cr |@\discretionary via @@3 b=44 p=50 d=15416|\cr |@@5: line 3.1- t=19241 -> @@3|\cr |ments.|\cr |@\par via @@4 b=0 p=-10000 d=5100|\cr |@\par via @@5 b=0 p=-10000 d=5100|\cr |@@6: line 4.2- t=24341 -> @@5|\cr \enddisplay Lines that begin with `|@@|' ^^{atsign atsign} represent {\sl^{feasible breakpoints}}, i.e., breakpoints that can be reached without any badness exceeding the tolerance. Feasible breakpoints are numbered consecutively, starting with |@@1|; the beginning of the paragraph is considered to be feasible too, and it is number |@@0|. Lines that begin with `|@|' but not `|@@|' are candidate ways to reach the feasible breakpoint that follows; \TeX\ will select only the best candidate, when there is a choice. Lines that do not begin with `|@|' indicate how far \TeX\ has gotten in the paragraph. Thus, for example, we find `|@@2: line 2.0 t=32481 -> @@1|' after `|...hap-pi-est when|' and before `|he|', so we know that feasible breakpoint~|@@2| occurs at the space between the words |when| and |he|. The notation `|line 2.0|' means that this feasible break comes at the end of line~2, and that this line will be very loose. \ (The suffixes |.0|, |.1|, |.2|, |.3| stand respectively for very loose, loose, decent, and tight.) \ A hyphen is suffixed to the line number if that line ends with a discretionary break, or if it is the final line of the paragraph; for example, `|line 1.2-|' is a decent line that was hyphenated. The notation `|t=32481|' means that the total demerits from the beginning of the paragraph to~|@@2| are 32481, and `|-> @@1|' means that the best way to get to |@@2| is to come from |@@1|. On the preceding line of trace data we see the calculations for a typeset line to this point from |@@1|: The badness is~131, the penalty is~0, hence there are 29881 demerits. Similarly, breakpoint |@@3| presents an alternative for the second line of the paragraph, obtained by breaking between `|he|' and `|was|'; this one makes the second line tight, and it has only 3825 demerits when the demerits of line~1 are added, so it appears that |@@3| will work much better than |@@2|. However, the next feasible breakpoint (|@@4|) occurs after `|doc-|', and the line from |@@2| to~|@@4| has only 12621 demerits, while the line from |@@3| to~|@@4| has a whopping 103101; therefore the best way to get from |@@0| to~|@@4| is via~|@@2|. If we regard demerits as distances, \TeX\ is finding the ``^{shortest paths}'' from |@@0| to each feasible breakpoint (using a variant of a well-known algorithm for shortest paths in an acyclic graph). Finally the end of the paragraph comes at breakpoint |@@6|, and the shortest path from |@@0| to~|@@6| represents the best sequence of breakpoints. Following the arrows back from~|@@6|, we deduce that~the best breaks in this particular paragraph go through |@@5|, |@@3|, and~|@@1|. \ddangerexercise Explain why there are 29881 demerits from |@@1| to |@@2|, and 12621 demerits from~|@@2| to~|@@4|. \answer $(10+131)^2+0^2+10000=29881$ and $(10+1)^2+50^2+10000=12621$. In both cases the ^|\adjdemerits| were added because the lines were visually incompatible (decent, then very loose, then decent); plain \TeX's values for ^|\linepenalty| and |\adjdemerits| were used. \ddanger If `|b=*|' ^^|*| appears in such trace data, it means that an infeasible breakpoint had to be chosen because there was no feasible way to keep total demerits small. \danger We still haven't discussed the special trick that allows the final line of a paragraph to be shorter than the others. Just before \TeX\ begins to choose breakpoints, it does two important things: \ (1)~If the final item of the current horizontal list is glue, ^^|\unskip| that glue is discarded. \ (The reason is that a blank space often gets into a token list just before ^|\par| or just before |$$|, and this blank space should not be part of the paragraph.) \ (2)~Three more items are put at the end of the current horizontal list: |\penalty10000| (which prohibits a line break); |\hskip\parfillskip| (which adds ``^{finishing glue}'' to the paragraph); and |\penalty-10000| (which forces the final break). Plain \TeX\ sets ^|\parfillskip||=0pt plus1fil|, so that the last line of each paragraph will be filled with white space if necessary; but other settings of\/ |\parfillskip| are appropriate in special applications. For example, the present paragraph ends flush with the right margin, because it was typeset with |\parfillskip=0pt|; the author didn't have to rewrite any of the text in order to make this possible, since a long paragraph generally allows so much flexibility that a line break can be forced at almost any point. You can have some fun playing with paragraphs, because the algorithm for line breaking occasionally appears to be clairvoyant. Just write paragraphs that are long enough.\parfillskip=0pt % the \danger macro makes this \parfillskip local! \dangerexercise Ben ^{User} decided to say `|\hfilneg\par|' at the end of a paragraph, intending that the negative stretchability of\/ ^|\hfilneg| would cancel with the |\parfillskip| of plain \TeX\null. Why didn't his bright idea work? ^^{paragraph, ending} \answer Because \TeX\ discards a glue item that occurs just before |\par|. Ben should have said, e.g., `|\hfilneg\ \par|'. \dangerexercise How can you set |\parfillskip| so that the last line of a paragraph has exactly as much white space at the right as the first line has indentation at the left?\nobreak\hskip\parindent\hfilneg\ \answer Just say |\parfillskip|\stretch|=|\stretch|\parindent|. Of course, \TeX\ will not be able to find appropriate line breaks unless each paragraph is sufficiently long or sufficiently lucky; but with an appropriate text, your output will be immaculately symmetrical.{\parfillskip=\parindent\par} \ddangerexercise Since \TeX\ reads an entire paragraph before it makes any decisions about line breaks, the computer's memory capacity might ^^{capacity exceeded} be exceeded if you are typesetting the works of some ^^{Joyce, James} ^{philosopher} or modernistic novelist who writes 200-line paragraphs. Suggest a way to cope with such authors. \answer Assuming that the author is deceased and/or set in his or her ways, the remedy is to insert `|{\parfillskip=0pt\par\parskip=0pt\noindent}|' in random places, after each 50 lines or so of text. \ (Every space between words is usually a feasible breakpoint, when you get sufficiently far from the beginning of a paragraph.) \danger \TeX\ has two parameters called ^|\leftskip| and ^|\rightskip| that specify glue to be inserted at the left and right of every line in a paragraph; this glue is taken into account when badnesses and demerits are computed. Plain \TeX\ normally keeps |\leftskip| and |\rightskip| zero, but it has a `^|\narrower|' macro that increases both of their values by the current ^|\parindent|. You may want to use |\narrower| when ^{quoting} lengthy passages from a book. \begintt {\narrower\smallskip\noindent This paragraph will have narrower lines than the surrounding paragraphs do, because it uses the ``narrower'' feature of plain \TeX. The former margins will be restored after this group ends.\smallskip} \endtt (Try it.) \ The second `^|\smallskip|' in this example ends the paragraph. It's important to end the paragraph before ending the group, for otherwise the effect of\/ |\narrower| will disappear before \TeX\ begins to choose line breaks. \dangerexercise When an entire paragraph is typeset in ^{italic} or ^{slanted} type, it sometimes appears to be offset on the page with respect to other paragraphs. Explain how you could use |\leftskip| and |\rightskip| to shift all lines of a paragraph left by $1\pt$. \answer |{\leftskip=-1pt \rightskip=1pt| \ |\par}|\par \nobreak\medskip\noindent (This applies to a full paragraph; if you want to correct only isolated lines, you have to do it by hand.) \dangerexercise The ^|\centerline|, ^|\leftline|, ^|\rightline|, and ^|\line| macros of plain \TeX\ don't take |\leftskip| and |\rightskip| into account. How could you make them do so? \answer `|\def\line#1{\hbox to\hsize{\hskip\leftskip#1\hskip\rightskip}}|' is the only change needed. \ (Incidentally, ^{displayed equations} don't take account of\/ |\leftskip| and |\rightskip| either; it's more difficult to change that, because so many variations are possible.) \ddanger If you suspect that ^|\raggedright| setting is accomplished by some appropriate manipulation of\/ |\rightskip|, you are correct. But some care is necessary. For example, a person can set |\rightskip=0pt plus1fil|, and every line will be filled with space at the right. But this isn't a particularly good way to make ragged-right margins, because the infinite stretchability will assign zero badness to lines that are very short. To do a decent job of ragged-right setting, the trick is to set |\rightskip| so that it will stretch enough to make line breaks possible, yet not too much, because short lines should be considered bad. Furthermore the spaces between words should be fixed so that they do not stretch or shrink. \ (See the definition of\/ |\raggedright| in Appendix~B\null.) \ It would also be possible to allow a little variability in the interword glue, so that the right margin would not be quite so ragged but the paragraphs would still have an informal appearance. \danger \TeX\ looks at the parameters that affect line breaking only when it is breaking lines. For example, you shouldn't try to change the ^|\hyphenpenalty| in the middle of a paragraph, if you want \TeX\ to penalize the hyphens in one word more than it does in another word. The relevant values of |\hyphenpenalty|, |\rightskip|, |\hsize|, and so on, are the ones that are current at the end of the paragraph. On the other hand, the width of indentation that you get implicitly at the beginning of a paragraph or when you say `^|\indent|' is determined by the value of\/ ^|\parindent| at the time the indentation is contributed to the current horizontal list, not by its value at the end of the paragraph. Similarly, penalties that are inserted into math formulas within a paragraph are based on the values of\/ ^|\binoppenalty| and ^|\relpenalty| that are current at the end of each particular formula. Appendix~D contains an example that shows how to have both ragged-right and ragged-left margins within a single paragraph, without using |\leftskip| or |\rightskip|. \varunit=0.989pt % getting ready to make circular insert % \varunit=1.078pt was used with amr5: it had more letterspacing \setbox0=\vtop{\null \baselineskip6\varunit \parfillskip0pt \parshape 19 -18.25\varunit 36.50\varunit -30.74\varunit 61.48\varunit -38.54\varunit 77.07\varunit -44.19\varunit 88.39\varunit -48.47\varunit 96.93\varunit -51.70\varunit 103.40\varunit -54.08\varunit 108.17\varunit -55.72\varunit 111.45\varunit -56.68\varunit 113.37\varunit -57.00\varunit 114.00\varunit -56.68\varunit 113.37\varunit -55.72\varunit 111.45\varunit -54.08\varunit 108.17\varunit -51.70\varunit 103.40\varunit -48.47\varunit 96.93\varunit -44.19\varunit 88.39\varunit -38.54\varunit 77.07\varunit -30.74\varunit 61.48\varunit -18.25\varunit 36.50\varunit \fiverm \frenchspacing \noindent \hbadness 6000 \tolerance 9999 \pretolerance 0 \hyphenation{iso-peri-met-ric} The area of a circle is a mean proportional between any two regular and similar polygons of which one circumscribes it and the other is isoperimetric with it. In addition, the area of the circle is less than that of any circumscribed polygon and greater than that of any isoperimetric polygon. And further, of these circumscribed polygons, the one that has the greater number of sides has a smaller area than the one that has a lesser number; but, on the other hand, the isoperimetric polygon that has the greater number of sides is the larger. \hbox to 36.50\varunit{\hss[Galileo,\thinspace1638]\hss} } \danger \parshape 16 3pc 26pc 3pc 26pc 0pc 24.69pc 0pc 23.51pc 0pc 22.73pc 0pc 22.20pc 0pc 21.85pc 0pc 21.65pc 0pc 21.58pc 0pc 21.65pc 0pc 21.85pc 0pc 22.20pc 0pc 22.73pc 0pc 23.51pc 0pc 24.69pc 0pc 29pc \vadjust{\moveright 28pc\vbox to 0pt{\vskip88pt\vskip-60\varunit \vskip-3pt\box0\vss}}% \strut It's possible to control the length of lines in a much more general way, if simple changes to |\leftskip| and |\rightskip| aren't flexible enough for your purposes. For example, a semicircular ^{hole} has been cut out of the present paragraph, in order to make room for a circular illustration that contains some of ^{Galileo}'s immortal words about ^{circle}s; all of the line breaks in this paragraph and in the circular quotation were found by \TeX's line-breaking algorithm. You can specify an essentially arbitrary paragraph shape by saying ^|\parshape||=|\, where the \ is a positive integer $n$, followed by $2n$ \ specifications. In general, `|\parshape=|$n$ $i_1$~$l_1$ $i_2$~$l_2$ $\ldots$ $i_n$~$l_n$' specifies a paragraph whose first $n$ lines will have lengths $l_1$, $l_2$, \dots,~$l_n$, respectively, and they will be indented from the left margin by the respective amounts $i_1$, $i_2$, \dots,~$i_n$. If the paragraph has fewer than $n$ lines, the additional specifications will be ignored; if it has more than $n$ lines, the specifications for line $n$ will be repeated ad infinitum. You can cancel the effect of a previously specified |\parshape| by saying `|\parshape=0|'.\parfillskip0pt ^^{illustrations, fitting copy around} \ddangerexercise Typeset the following ^{Pascal}ian quotation in the shape of an isosceles ^{triangle}: ``I turn, in the following treatises, to various uses of those triangles whose generator is unity. But I leave out many more than I include; it is extraordinary how fertile in properties this triangle is. Everyone can try his hand.'' \answer The author's best solution is based on a variable |\dimen| register |\x|: \begintt \setbox1=\hbox{I} \setbox0=\vbox{\parshape=11 -0\x0\x -1\x2\x -2\x4\x -3\x6\x -4\x8\x -5\x10\x -6\x12\x -7\x14\x -8\x16\x -9\x18\x -10\x20\x \ifdim \x>2em \rightskip=-\wd1 \else \frenchspacing \rightskip=-\wd1 plus1pt minus1pt \leftskip=0pt plus 1pt minus1pt \fi \parfillskip=0pt \tolerance=1000 \noindent I turn, ... hand.} \centerline{\hbox to \wd1{\box0\hss}} \endtt Satisfactory results are obtained with font |cmr10| when |\x| is set to $8.9\pt$, $13.4\pt$, $18.1\pt$, $22.6\pt$, $32.6\pt$, and $47.2\pt$, yielding triangles that are respectively 11,~9, 8, 7, 6, and~5 lines tall. \danger You probably won't need unusual parshapes very often. But there's a special case that occurs rather frequently, so \TeX\ provides a special abbreviation for~it in terms of two parameters called ^|\hangindent| and ^|\hangafter|. The command `|\hangindent=|\' specifies a so-called ^{hanging indentation}, and the command `|\hangafter=|\' specifies the duration of that indentation. Let $x$ and $n$ be the respective values of\/ |\hangindent| and |\hangafter|, and let $h$ be the value of\/ ^|\hsize|; then if $n\ge0$, hanging indentation will occur on lines $n+1$, $n+2$, $\ldots$ of the paragraph, but if $n<0$ it will occur on lines 1,~2, \dots,~$\vert n\vert$. Hanging indentation means that lines will be of width $h-\vert x\vert$ instead of their normal width~$h$; if $x\ge0$, the lines will be indented at the left margin, otherwise they will be indented at the right margin. For example, the ``dangerous bend'' paragraphs of this manual have a hanging indentation of 3~picas that lasts for two lines; they were set with |\hangindent=3pc| and |\hangafter=-2|. \danger Plain \TeX\ uses hanging indentation in its `^|\item|' macro, which produces a paragraph in which every line has the same indentation as a normal |\indent|. Furthermore, |\item| takes a parameter that is placed into the position of the indentation on the first line. Another macro called `^|\itemitem|' does the same thing but with double indentation. For example, suppose you type \begintt \item{1.} This is the first of several cases that are being enumerated, with hanging indentation applied to entire paragraphs. \itemitem{a)} This is the first subcase. \itemitem{b)} And this is the second subcase. Notice that subcases have twice as much hanging indentation. \item{2.} The second case is similar. \endtt {\let\par=\endgraf Then you get the following output: \medskip \item{1.} This is the first of several cases that are being enumerated, with hanging indentation applied to entire paragraphs. \itemitem{a)} This is the first subcase. \itemitem{b)} And this is the second subcase. Notice that subcases have twice as much hanging indentation. \item{2.} The second case is similar. \medskip}\noindent\hangindent0pt (Indentations in plain \TeX\ are not actually as dramatic as those displayed here; Appendix~B says `|\parindent=20pt|', but this manual has been set with |\parindent=36pt|.) \ It is customary to put ^|\medskip| before and after a group of itemized paragraphs, and to say |\noindent| before any closing remarks that apply to all of the cases. ^^{enumerated cases in separate paragraphs} Blank lines are not needed before |\item| or |\itemitem|, since those macros begin with |\par|. \dangerexercise Suppose one of the enumerated cases continues for two or more paragraphs. How can you use |\item| to get hanging indentation on the subsequent paragraphs? \answer |\item{}| at the beginning of each paragraph that wants hanging indentation. \dangerexercise Explain how to make a ``^{bullet}ed'' item that says `$\bullet$' instead of `1.'. \answer |\item{$\bullet$}| \ddangerexercise The `|\item|' macro doesn't alter the right-hand margin. How could you indent at both sides? \answer Either change |\hsize| or |\rightskip|. The trick is to change it back again at the end of a paragraph. Here's one way, without grouping: \begintt \let\endgraf=\par \edef\restorehsize{\hsize=\the\hsize} \def\par{\endgraf \restorehsize \let\par=\endgraf} \advance\hsize by-\parindent \endtt \ddangerexercise Explain how you could specify a hanging indentation of $-2$ ems (i.e., the lines should project into the left margin), after the first two lines of a paragraph. \answer |\dimen0=\hsize \advance\dimen0 by 2em|\parbreak |\parshape=3 0pt\hsize 0pt\hsize -2em\dimen0| \danger If\/ |\parshape| and hanging indentation have both been specified, |\parshape| takes precedence and |\hangindent| is ignored. You get the normal paragraph shape, in which every line width is |\hsize|, when |\parshape=0|, |\hangindent=0pt|, and |\hangafter=1|. \TeX\ automatically restores these normal values at the end of every paragraph, and (by local definitions) whenever it enters internal vertical mode. For example, hanging indentation that might be present outside of a ^|\vbox| construction won't occur inside that vbox, unless you ask for it inside. ^^{paragraph shape reset} ^^{hanging indentation reset} \ddangerexercise Suppose you want to leave room at the right margin for a rectangular illustration that takes up 15 lines, and you expect that three paragraphs will go by before you have typeset enough text to get past that illustration. Suggest a good way to do this without trial and error, given the fact that \TeX\ resets hanging indentation. \answer The three paragraphs can be combined into a single paragraph, if you use `|\hfil\vadjust{\vskip\parskip}\break\indent|' instead of `|\par|' after the first two. Then of course you say, e.g., |\hangindent=-50pt \hangafter=-15|. \ (The same idea can be applied in connection with |\looseness|, if you want \TeX\ to make one of three paragraphs looser but if you don't want to choose which one it will be. However, long paragraphs fill \TeX's memory; please use restraint.) \ See also the next exercise. \ddanger If ^{displayed equations} occur in a paragraph that has a nonstandard shape, \TeX\ always assumes that the display takes up exactly three lines. For example, a paragraph that has four lines of text, then a display, then two more lines of text, is considered to be $4+3+2=9$ lines long; the displayed equation will be indented and centered using the paragraph shape information appropriate to line~6. \ddanger \TeX\ has an internal integer variable called ^|\prevgraf| that records the number of lines in the most recent paragraph that has been completed or partially completed. You can use |\prevgraf| in the context of a \, and you can set |\prevgraf| to any desired nonnegative value if you want to make \TeX\ think that it is in some particular part of the current paragraph shape. For example, let's consider again a paragraph that contains four lines plus a display plus two more lines. When \TeX\ starts the paragraph, it sets |\prevgraf=0|; when it starts the display, |\prevgraf| will be~4; when it finishes the display, |\prevgraf| will be~7; and when it ends the paragraph, |\prevgraf| will be~9. If the display is actually one line taller than usual, you could set |\prevgraf=8| at the beginning of the two final lines; then \TeX\ will think that a 10-line paragraph is being made. The value of\/ |\prevgraf| affects line breaking only when \TeX\ is dealing with nonstandard |\parshape| or |\hangindent|. \edef\lastex{\chapno.\the\exno} \ddangerexercise Solve exercise \lastex\ using |\prevgraf|. \answer Use |\hangcarryover| between paragraphs, defined as follows: \begintt \def\hangcarryover{\edef\next{\hangafter=\the\hangafter \hangindent=\the\hangindent} \par\next \edef\next{\prevgraf=\the\prevgraf} \indent\next} \endtt \ddanger You are probably convinced by now that \TeX's line-breaking algorithm has plenty of bells and whistles, perhaps even too many. But there's one more feature, called ``looseness''; some day you might find yourself needing it, when you are fine-tuning the pages of a book. If you set |\looseness=1|, \TeX\ will try to make the current paragraph one line longer than its optimum length, provided that there is a way to choose such breakpoints without exceeding the tolerance you have specified for the badnesses of individual lines. Similarly, if you set |\looseness=2|, \TeX\ will try to make the paragraph two lines longer; and |\looseness=-1| causes an attempt to make it shorter. The general idea is that \TeX\ first finds breakpoints as usual; then if the optimum breakpoints produce $n$~lines, and if the current ^|\looseness| is~$l$, \TeX\ will choose the final breakpoints so as to make the final number of lines as close as possible to $n+l$ without exceeding the current tolerance. Furthermore, the final breakpoints will have fewest total demerits, considering all ways to achieve the same number of~lines. \ddanger For example, you can set |\looseness=1| if you want to avoid a lonely ``^{club line}'' or ``^{widow line}'' on some page that does not have sufficiently flexible glue, or if you want the total number of lines in some two-column document to come out to be an even number. It's usually best to choose a paragraph that is already pretty ``full,'' i.e., one whose last line doesn't have much white space, since such paragraphs can generally be loosened without much harm. You might also want to insert a ^{tie} between the last two words of that paragraph, so that the loosened version will not end with only one ``^{widow word}'' on the ^^{orphans, see widow words} line; this tie will cover your tracks, so that people will find it hard to detect the fact that you have tampered with the spacing. On the other hand, \TeX\ can take almost any sufficiently long paragraph and stretch it a bit, without substantial harm; the present paragraph is, in fact, one line looser than \hbox{its optimum length}.\looseness=1 \ddanger \TeX\ resets the looseness to zero at the same time as it resets |\hangindent|, |\hangafter|, and |\parshape|. \ddangerexercise Explain what \TeX\ will do if you set |\looseness=-1000|. \answer It will set the current paragraph in the minimum number of lines that can be achieved without violating the tolerance; and, given that number of lines, it will break them optimally. \ (However, nonzero looseness makes \TeX\ work harder, so this is not recommended if you don't want to pay for the extra computation. You can achieve almost the same result much more efficiently by setting ^|\linepenalty||=100|, say.) \danger Just before switching to horizontal mode to begin scanning a paragraph, \TeX\ inserts the glue specified by ^|\parskip| into the vertical list that will contain the paragraph, unless that vertical list is empty so far. For example, `|\parskip=3pt|' will cause 3~points of extra space to be placed between paragraphs. Plain \TeX\ sets |\parskip=0pt plus1pt|; this gives a little stretchability, but no extra space. \danger After line breaking is complete, \TeX\ appends the lines to the current vertical list that encloses the current paragraph, inserting interline glue as explained in Chapter~12; this interline glue will depend on the values of\/ ^|\baselineskip|, ^|\lineskip|, and ^|\lineskiplimit| that are currently in force. \TeX\ will also insert penalties into the vertical list, just before each glob of ^{interline glue}, in order to help control page breaks that might have to be made later. For example, a special penalty will be assessed for breaking a page between the first two lines of a paragraph, or just before the last line, so that ``club'' or ``widow'' lines that are detached from the rest of a paragraph will not appear all alone on a page unless the alternative is worse. \danger Here's how interline penalties are calculated: \TeX\ has just chosen the breakpoints for some paragraph, or for some partial paragraph that precedes a displayed equation; and $n$~lines have been formed. The penalty between lines $j$ and $j+1$, given a value of $j$ in the range $1\le j|}| within a paragraph, \TeX\ will insert the specified internal vertical list into the vertical list that encloses the paragraph, immediately after whatever line contained the position of the |\vadjust|. For example, you can say `|\vadjust{\kern1pt}|' to increase the amount of space between lines of a paragraph if those lines would otherwise come out too close together. \ (The \vadjust{\vskip1pt}author did it in the previous line, just to illustrate what happens.) \ Also, if you want to make sure that a page break will occur immediately after a certain line, you can say `|\vadjust{\eject}|' ^^|\eject| anywhere in that line. \ddanger Later chapters discuss |\insert| and |\mark| commands that are relevant to \TeX's page builder. If such commands appear within a paragraph, they are removed from whatever horizontal lines contain them and placed into the enclosing vertical list, together with other vertical material from |\vadjust| commands that might be present. In the final vertical list, each horizontal line of text is an hbox that is immediately preceded by interline glue and immediately followed by vertical material that has ``^{migrate}d out'' from that line (with left to right order preserved, if there are several instances of vertical material); then comes the interline penalty, if it is nonzero. Inserted vertical material does not influence the ^{interline glue}. \ddangerexercise Design a |\marginalstar| macro ^^{marginal notes} that can be used anywhere in a paragraph. It should use |\vadjust| to place an asterisk in the margin just to the left of the line where |\marginalstar| occurs. \answer The tricky part is to avoid ``opening up'' the paragraph by adding anything to its height; yet this star is to be contributed after a line having an unknown depth, because the depth of the line depends on details of line breaking that aren't known until afterwards. The following solution uses ^|\strut|, and assumes that the line containing the marginal star does not have depth exceeding ^|\dp||\strutbox|, the depth of a ^|\strut|. \begintt \def\strutdepth{\dp\strutbox} \def\marginalstar{\strut\vadjust{\kern-\strutdepth\specialstar}} \endtt Here |\specialstar| is a box of height zero and depth |\strutdepth|, and it puts an asterisk in the left margin: \begintt \def\specialstar{\vtop to \strutdepth{ \baselineskip\strutdepth \vss\llap{* }\null}} \endtt \ddanger When \TeX\ enters ^{horizontal mode}, it will interrupt its normal scanning to read tokens that were predefined by the command ^|\everypar||={|\|}|. For example, suppose you have said `|\everypar={A}|'. If you type `|B|' in vertical mode, \TeX\ will shift to horizontal mode (after contributing ^|\parskip| glue to the current page), and a horizontal list will be initiated by inserting an empty box of width ^|\parindent|. Then \TeX\ will read `|AB|', since it reads the |\everypar| tokens before getting back to the `|B|' that triggered the new paragraph. Of course, this is not a very useful illustration of |\everypar|; but if you let your imagination run you will think of better applications. \ddangerexercise Use |\everypar| to define an |\insertbullets| macro: All paragraphs in a group of the form `|{\insertbullets ...\par}|' should have a bullet symbol `$\bullet$' as part of their indentation. ^^{bulleted lists} \answer |\def\insertbullets{\everypar={\llap{$\bullet$\enspace}}}|\par \nobreak\smallskip\noindent (A similar device can be used to insert hanging indentation, and/or to number the paragraphs automatically.) \ddanger A paragraph of zero lines is formed if you say `|\noindent\par|'. If\/ |\everypar| is null, such a paragraph contributes nothing except |\parskip| glue to the current vertical list. \ddangerexercise Guess what happens if you say `|\noindent$$...$$ \par|'. \answer First comes |\parskip| glue (but you might not see it on the current page if you say |\showlists|, since glue disappears at the top of each page). Then comes the result of\/ |\everypar|, but let's assume that |\everypar| doesn't add anything to the horizontal list, so that you get an empty horizontal list; then there's no partial paragraph before the display. The displayed equation follows the normal rules (it occupies lines 1--3 of the paragraph, and uses the indentation and length of line~2, if there's a nonstandard shape). Nothing follows the display, since a blank space is ignored after a closing `|$$|'.\par Incidentally, the behavior is different if you start a paragraph with `|$$|' instead of with |\noindent$$|, ^^{display at beginning of paragraph} since \TeX\ inserts a paragraph indentation that will appear on a line by itself (with |\leftskip| and |\parfillskip| and |\rightskip| glue). \ddanger Experience has shown that \TeX's line-breaking algorithm can be harnessed to a surprising variety of tasks. Here, for example, is an application that indicates one of the possibilities: Articles that are published in {\sl^{Mathematical Reviews}\/\null} are generally signed with the reviewer's name and address, and this information is typeset flush right, i.e., at the right-hand margin. ^^{flush right} If there is sufficient space to put such a name and address at the right of the final line of the paragraph, the publishers can save space, and at the same time the results look better because there are no strange gaps on the page. \def\signed #1 (#2){{\unskip\nobreak\hfil\penalty50\hskip2em \hbox{}\nobreak\hfil\sl#1\/ \rm(#2) \parfillskip=0pt \finalhyphendemerits=0 \endgraf}} \begindisplay \vbox{\hsize 3.0in \parindent0pt This is a case where the name and address fit in nicely with the review. \signed A. Reviewer (Ann Arbor, Mich.) \medskip But sometimes an extra line must be added. \signed N. Bourbaki (Paris)} \enddisplay ^^{Reviewer} ^^{Bourbaki} Let's suppose that a space of at least two ems should separate the reviewer's name from the text of the review, if they occur on the same line. We would like to design a macro so that the examples shown above could be typed as follows in an input file: \begintt ... with the review. \signed A. Reviewer (Ann Arbor, Mich.) ... an extra line must be added. \signed N. Bourbaki (Paris) \endtt Here is one way to solve the problem: \begintt \def\signed #1 (#2){{\unskip\nobreak\hfil\penalty50 \hskip2em\hbox{}\nobreak\hfil\sl#1\/ \rm(#2) \parfillskip=0pt \finalhyphendemerits=0 \par}} \endtt If a line break occurs at the |\penalty50|, the |\hskip2em| will disappear and the empty |\hbox| will occur at the beginning of a line, followed by |\hfil| glue. This yields two lines whose badness is zero; the first of these lines is assessed a penalty of~50. But if no line break occurs at the |\penalty50|, there will be glue of $2\em$ plus $2\,{\rm fil}$ between the review and the name; this yields one line of badness zero. \TeX\ will try both alternatives, to see which leads to the fewest total demerits. The one-line solution will usually be preferred if it is feasible. \ddangerexercise Explain what would happen if `|\hbox{}|' were left out of the ^|\signed| macro. \answer A break at |\penalty50| would cancel |\hskip2em\nobreak\hfil|, so the next line would be forced to start with the reviewer's name flush left. \ (But ^|\vadjust||{}| would actually be better than |\hbox{}|; it uses \TeX\ more efficiently.) \ddangerexercise Why does the |\signed| macro say `^|\finalhyphendemerits||=0|'\thinspace? \answer Otherwise the line-breaking algorithm might prefer two final lines to one final line, simply in order to move a hyphen from the second-last line up to the third-last line where it doesn't cause demerits. This in fact caused some surprises when the |\signed| macro was being tested; |\tracingparagraphs=1| was used to diagnose the problem. {\hbadness=10000 \ddangerexercise In one of the paragraphs earlier in this chapter, the author used ^|\break| to force a line break in a specific place; as a result, the third line of that particular paragraph was really spaced out.\break Explain why all the extra space went into the third line, instead of being distributed impartially among the first three lines. \answer Distributing the extra space evenly would lead to three lines of the maximum badness (10000). It's better to have just one bad line instead of three, since \TeX\ doesn't distinguish degrees of badness when lines are really awful. In this particular case the ^|\tolerance| was 200, so \TeX\ didn't try any line breaks that would stretch the first two lines; but even if the tolerance had been raised to 10000, the optimum setting would have had only one underfull line. If you really want to spread the space evenly you can do so by using ^|\spaceskip| to increase the amount of stretchability between words. } \ddanger If you want to avoid overfull boxes at all costs without trying to fix them manually, you might be tempted to set |\tolerance=10000|; this allows arbitrarily bad lines to be acceptable in tough situations. But infinite tolerance is a bad idea, because \TeX\ doesn't distinguish between terribly bad and preposterously horrible lines. Indeed, a tolerance of 10000 encourages \TeX\ to concentrate all the badness in one place, making one truly unsightly line instead of two moderately bad ones, because a single ``write-off'' produces fewest total demerits according to the rules. There's a much better way to get the desired effect: \TeX\ has a parameter called ^|\emergencystretch| that is added to the assumed stretchability of every line when badness and demerits are computed, in cases where overfull boxes are otherwise unavoidable. If |\emergencystretch| is positive, \TeX\ will make a third pass over a paragraph before choosing the line breaks, when the first passes did not find a way to satisfy the ^|\pretolerance| and ^|\tolerance|. The effect of\/ |\emergencystretch| is to scale down the badnesses so that large infinities are distinguishable from smaller ones. By setting |\emergencystretch| high enough (based on |\hsize|) you can be sure that the |\tolerance| is never exceeded; hence overfull boxes will never occur unless the line-breaking task is truly impossible. \ddangerexercise Devise a ^|\raggedcenter| macro (analogous to ^|\raggedright|) that partitions the words of a paragraph into as few as possible lines of approximately equal size and centers each individual line. Hyphenation should be avoided if possible. \answer |\def\raggedcenter{\leftskip=0pt plus4em \rightskip=\leftskip|% \parbreak|\parfillskip=0pt \spaceskip=.3333em \xspaceskip=.5em|\parbreak |\pretolerance=9999 \tolerance=9999 \parindent=0pt|\parbreak |\hyphenpenalty=9999 \exhyphenpenalty=9999 }| \endchapter When the author objects to [a hyphenation]\/ he should be asked to add or cancel or substitute a word or words that will prevent the breakage. \smallskip Authors who insist on even spacing always, with sightly divisions always, do not clearly understand the rigidity of types. \author T. L. ^{DE VINNE}, {\sl Correct Composition\/} (1901) % p138, p206 \bigskip In reprinting his own works, whenever [William ^{Morris}]\/ found a line that justified awkwardly, he altered the wording solely for the sake of making it look well in print. \smallskip When a proof has been sent me with two or three lines so widely spaced as to make a grey band across the page, I have often rewritten the passage so as to fill up the lines better; but I am sorry to say that my object has generally been so little understood that the compositor has spoilt all the rest of the paragraph instead of mending his former bad work. \author GEORGE BERNARD ^{SHAW}, in {\sl The Dolphin\/} (1940) % v4 p80 \eject \beginchapter Chapter 15. How \TeX\ Makes\\Lines into Pages \tracingpages=1 \TeX\ attempts to choose desirable places to divide your document into individual pages, and its technique for doing this usually works pretty well. But the problem of ^{page make-up} is considerably more difficult than the problem of line breaking that we considered in the previous chapter, because pages often have much less flexibility than lines do. If the vertical glue on a page has little or no ability to stretch or to shrink, \TeX\ usually has no choice about where to start a new page; conversely, if there is too much variability in the glue, the result will look bad because different pages will be too irregular. Therefore if you are fussy about the appearance of pages, you can expect to do some rewriting of the manuscript until you achieve an appropriate balance, or you might need to fiddle with the ^|\looseness| as described in Chapter~14; no automated system will be able to do this as well as you. Mathematical papers that contain a lot of displayed equations have an advantage in this regard, because the glue that surrounds a display tends to be quite flexible. \TeX\ also gets valuable room to maneuver when you have occasion to use ^|\smallskip| or ^|\medskip| or ^|\bigskip| spacing between certain paragraphs. For example, consider a page that contains a dozen or so exercises, and suppose that there is $3\pt$ of additional space between exercises, where this space can stretch to $4\pt$ or shrink to $2\pt$. Then there is a chance to squeeze an extra line on the page, or to open up the page by removing one line, in order to avoid splitting an exercise between pages. Similarly, it is possible to use flexible glue in special publications like membership rosters or company telephone directories, so that individual entries need not be split between columns or pages, yet every column appears to be the same height. For ordinary purposes you will probably find that \TeX's automatic method of page breaking is satisfactory. And when it occasionally gives unpleasant results, you can force the machine to break at your favorite place by typing `^|\eject|'. But be careful: |\eject| will cause \TeX\ to stretch the page out, if necessary, so that the top and bottom baselines agree with those on other pages. If you want to eject a short page, filling it with blank space at the bottom, type `|\vfill\eject|' instead. \danger If you say `|\eject|' in the middle of a paragraph, the paragraph will end first, as if you typed `|\par\eject|'. But Chapter~14 mentions that you can say `^|\vadjust||{\eject}|' in mid-paragraph, if you want to force a page break after whatever line contains your current position when the full paragraph is eventually broken up into lines; the rest of the paragraph will go on the following page. \danger To prevent a page break, you can say `^|\nobreak|' in vertical mode, just as |\nobreak| in horizontal mode prevents breaks between lines. For example, it is wise to say |\nobreak| between the title of a subsection and the first line of text in that subsection. But |\nobreak| does not cancel the effect of other commands like |\eject| that tell \TeX\ to break; it only inhibits a break at glue that immediately follows. You should become familiar with \TeX's rules for line breaks and page breaks if you want to maintain fine control over everything. The remainder of this chapter is devoted to the intimate details of page breaking. \ninepoint \danger \TeX\ breaks lists of lines into pages by computing badness ratings and penalties, more or less as it does when breaking paragraphs into lines. But pages are made up one at a time and removed from \TeX's memory; there is no looking ahead to see how one page break will affect the next one. In other words, \TeX\ uses a special method to find the optimum breakpoints for the lines in an entire paragraph, but it doesn't attempt to find the optimum breakpoints for the pages in an entire document. The computer doesn't have enough high-speed memory capacity to remember the contents of several pages, so \TeX\ simply chooses each page break as best it can, by a process of ``local'' rather than ``global'' optimization. \tracingpages=0 \danger Let's look now at the details of \TeX's page-making process. Everything you contribute to the pages of your document is placed on the {\sl ^{main vertical list}}, which is the sequence of items that \TeX\ has accumulated while in vertical mode. Each item in a ^{vertical list} is one of the following types of things:\enddanger \smallskip \item\bull a box (an hbox or vbox or rule); \item\bull a ``^{whatsit}'' (something special to be explained later); \item\bull a ^{mark} (another thing that will be explained later); \item\bull an ^{insertion} (yet another thing that we will get to); \item\bull a glob of ^{glue} (or ^|\leaders|, as we will see later); \item\bull a ^{kern} (something like glue that doesn't stretch or shrink); \item\bull a ^{penalty} (representing the undesirability of breaking here). \smallskip\noindent The last three types (glue, kern, and penalty items) are called ^{discardable}, for the same reason that we called them discardable in horizontal lists. You might want to compare these specifications with the analogous rules for the horizontal case, found in Chapter~14; it turns out that vertical lists are just like horizontal ones except that character boxes, discretionary breaks, ^|\vadjust| items, and math shifts cannot appear in vertical lists. Chapter~12 exhibits a typical vertical list in \TeX's internal box-and-glue representation. \danger Page breaks can occur only at certain places within a vertical list. The permissible breakpoints are exactly the same as in the horizontal case, namely\enddanger \smallskip \item{a)} at glue, provided that this glue is immediately preceded by a non-discardable item (i.e., by a box, whatsit, mark, or insertion); \smallskip \item{b)} at a kern, provided that this kern is immediately followed by glue; \smallskip \item{c)} at a penalty (which might have been inserted automatically in a paragraph). \smallskip\noindent Interline glue is usually inserted automatically between the boxes of a vertical list, as explained in Chapter~12, so there is usually a valid breakpoint between boxes. \danger As in horizontal lists, each potential breakpoint has an associated penalty, which is high for undesirable breakpoints and negative for desirable ones. The penalty is zero at glue and kern breaks, so it is nonzero only at explicit penalty breaks. If you say `^|\penalty||-100|' between two paragraphs, you are indicating that \TeX\ should try to break here because the penalty is negative; a bonus of 100 points for breaking at this place will essentially cancel up to 100 units of badness that might be necessary to achieve such a break. A penalty of 10000 or more is so large that it inhibits breaking; a penalty of $-10000$ or less is so small that it forces breaking. \danger Plain \TeX\ provides several control sequences that help to control page breaks. For example, ^|\smallbreak|, ^|\medbreak|, and ^|\bigbreak| specify increasingly desirable places to break, having respective penalties of $-50$, $-100$, and~$-200$; furthermore, they will insert a ^|\smallskip|, ^|\medskip|, or ^|\bigskip| of space, respectively, if a break is not taken. However, |\smallbreak|, |\medbreak|, and |\bigbreak| do not increase existing glue unnecessarily; for example, if you say |\smallbreak| just after a displayed equation, you won't get a |\smallskip| of space in addition to the glue that already follows a display. Therefore these commands can conveniently be used before and after the statements of ^{theorems}, in a format for mathematical papers. In the present manual the author has used a macro that puts |\medbreak| before and after every dangerous-bend paragraph; |\medbreak\medbreak| is equivalent to a single |\medbreak|, so you don't see two medskips when one such paragraph ends and another one~begins. \danger The ^|\goodbreak| macro is an abbreviation for `|\par\penalty-500|'. This is a good thing to insert in your manuscript when proofreading, if you are willing to stretch some page a little bit extra in order to improve the following one. Later on if you make another change so that this |\goodbreak| command does not appear near the bottom of a page, it will have no effect; thus it is not as drastic as |\eject|. \danger The most interesting macro that plain \TeX\ provides for page make-up is called ^|\filbreak|. It means, roughly, ``Break the page here and fill the bottom with blank space, unless there is room for more copy that is itself followed by |\filbreak|.'' Thus if you put |\filbreak| at the end of every paragraph, and if your paragraphs aren't too long, every page break will occur between paragraphs, and \TeX\ will fit as many paragraphs as possible on each page. The precise meaning of\/ |\filbreak|~is \begintt \vfil\penalty-200\vfilneg \endtt according to Appendix B\null; and this simple combination of \TeX's primitives produces the desired result: If a break is taken at the |\penalty-200|, the preceding |\vfil| will fill the bottom of the page with blank space, and the ^|\vfilneg| will be discarded after the break; but if no break is taken at the penalty, the |\vfil| and |\vfilneg| will cancel each other and have no effect. \danger Plain \TeX\ also provides a ^|\raggedbottom| command, which is a vertical analog of\/ ^|\raggedright|: It tells \TeX\ to permit a small amount of variability in the bottom margins on different pages, in order to make the other spacing uniform. \ddanger We saw in Chapter 14 that breakpoints for paragraphs are chosen by computing ``demerits'' for each line and summing them over all lines. The situation for pages is simpler because each page is considered separately. \TeX\ figures the ``^{cost}'' of a page break by using the following formula: \begindisplay $\displaystyle{c=\cases{ p,&if $b<\infty$ and $p\le-10000$ and $q<10000$;\cr b+p+q,&if $b<10000$ and $-10000 values `^|\pagetotal|' and `^|\pagegoal|', respectively. You can even change them (but let's hope that you know what you are doing). For example, the command |\pagegoal=500pt| overrides the previously saved value of\/ |\vsize|. Besides |\pagetotal|, which represents the accumulated natural height, \TeX\ maintains the quantities ^|\pagestretch|, ^|\pagefilstretch|, ^|\pagefillstretch|, ^|\pagefilllstretch|, ^|\pageshrink|, and ^|\pagedepth|. When the current page contains no boxes, |\pagetotal| and its relatives are zero and |\pagegoal| is $16383.99998\pt$ (\TeX's largest \); changing their values has no effect at such times. The integer $q$ in the formula for page costs is also available for inspection and change; it is called ^|\insertpenalties|. \ddanger Page breaking differs from line breaking in one small respect that deserves mention here: If you say |\eject\eject|, the second |\eject| is ignored, because it is equivalent to |\penalty-10000| and penalties are discarded after a page break. But if you say |\break\break| in a paragraph, the second ^|\break| causes an empty line, because penalties are discarded after a break in a paragraph only if they do not belong to the final sequence of breakpoints. This technicality is unimportant in practice, because |\break\break| isn't a good way to make an empty line; that line will usually be an underfull hbox, since it has only the |\leftskip| and |\rightskip| glue in it. Similarly, `|\eject\eject|' would not be a good way to make an empty page, even if \TeX\ were to change its rules somehow so that an |\eject| would never be ignored. The best way to eject an ^{empty page} is to say `|\eject\line{}\vfil\eject|', and the best way to create an ^{empty line} is `|\break\hbox{}\hfil\break|'. Both of these avoid underfull boxes. \danger You are probably wondering how page numbers and such things get attached to pages. The answer is that \TeX\ allows you to do further processing after each page break has been chosen; a special ``output routine'' goes into action before pages actually receive their final form. Chapter~23 explains how to construct output routines and how to modify the output routine of plain~\TeX. \danger Every once in a~while, \TeX\ will produce a really awful-looking page and you will wonder what happened. For example, you might get just one paragraph and a lot of white space, when some of the text on the following page would easily fit into the white space. The reason for such apparently anomalous behavior is almost always that no good page break is possible; even the alternative that looks better to you is quite terrible as far as \TeX\ is concerned! \TeX\ does not distinguish between two choices that both have 10000 units of badness or more, even though some bad breaks do look much worse than others. The solution in such cases is to insert |\eject| or |\vfill\eject| in some acceptable spot, or to revise the manuscript. If this problem arises frequently, however, you probably are using a format that sets overly strict limitations on page format; try looking at the output of\/ |\tracingpages| and modifying some of \TeX's parameters, until you have better luck. \danger The remainder of this chapter is about insertions: things like footnotes and ^{illustrations}, and how they interact with page breaks. Before we discuss the primitive operations by which \TeX\ deals with insertions, we will take a look at the facilities that plain \TeX\ provides at a higher level. \danger Illustrations can be inserted in several ways using plain \TeX. The simplest of these is called a ``^{floating topinsert}''; you say \begindisplay ^|\topinsert|\^|\endinsert| \enddisplay and \TeX\ will attempt to put the vertical mode material at the top of the current page. If there's no room for such an insertion on this page, \TeX\ will insert it at the top of the next page. The \ can contain embedded paragraphs that temporarily interrupt vertical mode in the usual way; for example: \begintt \topinsert \vskip 2in \hsize=3in \raggedright \noindent{\bf Figure 3.} This is the caption to the third illustration of my paper. I have left two inches of space above the caption so that there will be room to introduce special artwork. \endinsert \endtt The ^{caption} in this example will be set ^{ragged-right} in a 3-inch column at the left of the page. Plain \TeX\ automatically adds a ``^{bigskip}'' below each topinsert; this will separate the caption from the text. The effects of\/ |\hsize=3in| and |\raggedright| do not extend past the |\endinsert|, since ^{grouping} is implied. \dangerexercise Modify this example so that the caption is moved over next to the right margin, instead of appearing at the left. \answer |\topinsert\vskip2in\rightline{\vbox{\hsize|\stretch|...|\stretch |artwork.}}\endinsert| does the job. But it's slightly more efficient to avoid ^|\rightline| by changing ^|\leftskip| as follows: `|\leftskip=\hsize \advance\leftskip by-3in|'. Then \TeX\ doesn't have to read the text of the caption twice. \danger Similarly, if you say `^|\pageinsert|\stretch\\stretch |\endinsert|', the vertical mode material will be justified to the size of a full page (without a bigskip below it); the result will appear on the following page. \danger There's also `^|\midinsert|\stretch\\stretch |\endinsert|', which tries first to insert the material in place, wherever you happen to be, in the middle of the current page. If there is enough room, you get the effect of^^|\bigskip|^^|\bigbreak| \begindisplay |\bigskip\vbox{|\|}\bigbreak| \enddisplay otherwise the |\midinsert| is effectively converted to a |\topinsert|. There is a slight probability that |\midinsert| will not find the best placement, because \TeX\ is sometimes processing text ahead of the current page. You may want to say `^|\goodbreak|' just before |\midinsert|. \danger You should use the commands |\topinsert|, |\pageinsert|, |\midinsert| in vertical mode (i.e., between paragraphs), not inside of boxes or other insertions. \danger If you have two or more |\topinsert| or |\pageinsert| commands in quick succession, \TeX\ may need to carry them over to several subsequent pages; but they will retain their relative order when they are carried over. For example, suppose you have pages that are nine inches tall, and suppose you have already specified 4~inches of text for some page, say page~25. Then suppose you make seven topinserts in a row, of respective sizes $1,2,3,9,3,2,1$ inches; the 9-inch one is actually a |\pageinsert|. What happens? Well, the first and second will appear at the top of page 25, followed by the 4~inches of copy you have already typed; that copy will immediately be followed by two more inches that you type after the seven inserts. The third topinsert will appear at the top of page~26, followed by six more inches of text; the fourth will fill page~27; and the remaining three will appear at the top of page~28. \dangerexercise What would happen in the example just discussed if the final 1-inch insertion were a |\midinsert| instead of a |\topinsert|? \answer It would appear on page~25, since it does fit there. A |\midinsert| will jump ahead of other insertions only if it is not carried over to another page; for example, if the second 3-inch insertion were a |\midinsert|, it would not appear on page~26, because it is converted to a |\topinsert| as soon as the |\midinsert| macro notices that the insertion is too big for page~25. \danger At the end of a paper, you probably want to make sure that no insertions are lost; and at the end of a chapter, you probably want to make sure that no insertions float into the following chapter. Plain \TeX\ will flush out all remaining insertions, with blank space filling the bottom of incomplete pages, if you say `|\vfill|^|\supereject|'. \danger Besides illustrations that are inserted at the top of a page, plain \TeX\ will also insert ^{footnotes} at the bottom of a page. The ^|\footnote| macro is provided for use within paragraphs;\footnote*{Like this.} for example, the footnote in the present sentence was typed in the following way: \begintt ... paragraphs;\footnote*{Like this.} for example, ... \endtt There are two parameters to a |\footnote|; first comes the ^{reference mark}, which will appear both in the paragraph\footnote{**}{The author typed `|paragraph\footnote{**}{The author ...}|' here.} and in the footnote itself, and then comes the text of the footnote.\footnote{$^{45}$}{And `|footnote.\footnote{$^{45}$}{And ...}|' here. The footnotes in this manual appear in smaller type, and they are set with hanging indentation; furthermore a smallskip occurs between footnotes on the same page. But in plain \TeX, footnotes are typeset with the normal size of type, with |\textindent| used for the reference mark, and without extra smallskips. The |\textindent| macro is like |\item|, but it omits hanging indentation.} The latter text may be several paragraphs long, and it may contain displayed equations and such things, but it should not involve other ^^|\textindent| ^^|\item| % these are in a split footnote, after the split! insertions. \TeX\ will ensure that each footnote occurs at the bottom of the same page as its reference.\footnote\dag{Printers often use the symbols |\dag| (\dag), |\ddag| (\ddag), |\S|~(\S), and |\P|~(\P) as reference marks; sometimes also |$\|\||$| ($\Vert$). You can say, e.g., `|\footnote\dag{...}|'.} A long footnote will be split, if necessary, and continued at the bottom of the following page, as you can see in the ^^|\dag|^^|\ddag|^^|\S|^^|\P|^^|\Vert| somewhat contrived example that appears here. Authors who are interested in good exposition should avoid footnotes whenever possible, since footnotes tend to be ^^{Gibbon} distracting.\footnote\ddag{Yet Gibbon's {\sl Decline and Fall\/} would not have been the same without footnotes.} \danger The |\footnote| macro should be used only in paragraphs or hboxes that are contributed to \TeX's main vertical list; insertions will be lost if they occur inside of boxes that are inside of boxes. Thus, for example, you should not try to put a |\footnote| into a subformula of a math formula. But it's OK to use footnotes within ^|\centerline|, e.g., ^^{Thor} \begintt \centerline{A paper by A. U. Thor% \footnote*{Supported by NSF.}} \endtt or even on the outer level of a table entry inside an ^|\halign|. \ddanger Topinserts work fine by themselves, and footnotes work fine by themselves, but complications can arise when you try to mix them in devious ways. For example, if a |\pageinsert| floats to the page that follows a long footnote that had to be broken, both of the held-over insertions may try to force themselves onto the same page, and an overfull vbox may result. Furthermore, insertions cannot appear within insertions, so you can't use |\footnote| within a |\topinsert|. If you really need a footnote in some caption, there's a ^|\vfootnote| macro that can be used in vertical mode. To use it, you put a reference mark like~`|*|' in the caption, and then you say `|\vfootnote*{The footnote}|' somewhere on the page where you guess that the caption will finally fall. In such complex circumstances you might want to rethink whether or not you are really using the most appropriate format for the exposition of your ideas. \ddanger Chapter 24 explains the exact rules about ^{migration} of vertical-mode material (like footnotes) from horizontal lists to the enclosing vertical list. Insertions, marks, and the results of\/ ^|\vadjust| all migrate in the same fashion. \danger Now let's study the primitives of \TeX\ that are used to construct macros like |\topinsert| and |\footnote|. We are about to enter behind the scenes into a sublanguage of \TeX\ that permits users to do complex manipulations with boxes and glue. Our discussion will be in two parts: First we shall consider \TeX's ``^{registers},'' with which a user can do ^{arithmetic} related to typesetting; and then we shall discuss the insertion items that can appear in horizontal and vertical lists. Our discussion of the first topic (registers) will be marked with single dangerous-bend signs, since registers are of general use in advanced applications of \TeX, whether or not they relate to insertions. But the second topic will be marked with double dangerous-bend signs, since insertions are rather esoteric. \danger \TeX\ has 256 registers called ^|\count||0| to |\count255|, each capable of containing integers between $-2147483647$ and $+2147483647$, inclusive; ^^\ i.e., the magnitudes should be less than $2^{31}$. \TeX\ also has 256 registers called ^|\dimen||0| to |\dimen255|, each capable of containing a ^\ (see Chapter~10). There are another 256 registers called ^|\skip||0| to |\skip255|, each containing ^\ (see Chapter~12); and ^|\muskip||0| to |\muskip255|, each containing ^\ (see Chapter~18). You can assign new values to these registers by saying \begindisplay |\count|\ |=| \\cr |\dimen|\ |=| \\cr |\skip|\ |=| \\cr |\muskip|\ |=| \\cr \enddisplay and then you can add or subtract values of the same type by saying^^|\advance| \begindisplay |\advance\count|\ |by| \\cr |\advance\dimen|\ |by| \\cr |\advance\skip|\ |by| \\cr |\advance\muskip|\ |by| \\cr \enddisplay For example, `|\dimen8=\hsize \advance\dimen8 by 1in|' sets register |\dimen8| to an inch more than the current value of the normal line size. \danger If infinite glue components are added, lower order infinities disappear. For example, after the two commands \begintt \skip2 = 0pt plus 2fill minus 3fill \advance\skip2 by 4pt plus 1fil minus 2filll \endtt the value of\/ |\skip2| will be $4\pt$ plus $2\,{\rm fill}$ minus $2\,{\rm filll}$. \danger Multiplication and division are possible too, but only by integers. For example, `^|\multiply||\dimen4 by 3|' triples the value of\/ |\dimen4|, and `^|\divide||\skip5 by 2|' cuts in half all three components of the glue that is currently registered in |\skip5|. You shouldn't divide by zero, nor should you multiply by numbers that will make the results exceed the register capacities. Division of a positive integer by a positive integer discards the remainder, and the sign of the result changes if you change the sign of either operand. For example, 14~divided by~3 yields~4; $-14$~divided by~3 yields~$-4$; $-14$~divided by~$-3$ yields~4. Dimension values are integer multiples of~^{sp} (scaled points). \danger You can use any |\count| register in the context of a ^\, any |\dimen| register in the context of a ^\, any |\skip| register in the context of ^\, and any |\muskip| register in the context of ^\. For example, `|\hskip\skip1|' puts horizontal glue into a list, using the value of\/ |\skip1|; and if\/ |\count5| is 20, the~command `|\advance\dimen20 by\dimen\count5|' is equivalent to `|\multiply\dimen20 by 2|'. \danger A |\dimen| register can be used also in the context of a \, and a |\skip| register can be used as a \ or a \. \TeX\ converts \ to \ by omitting the stretch and shrink components, and it converts \ to \ by assuming units of~sp (scaled points). For example, if\/ |\skip1| holds the value $1\pt$ plus~$2\pt$, then `|\dimen1=\skip1|' sets |\dimen1| equal to~$1\pt$; and the commands `|\count2=\dimen1|' or `|\count2=\skip1|' will set |\count2| equal to~65536. These rules also apply to \TeX's internal parameters; for example, `|\dimen2=\baselineskip|' will set |\dimen2| to the natural space component of the current baselineskip glue. \dangerexercise Test your knowledge of \TeX's registers by stating the results of each of the following commands when they are performed in sequence: \begintt \count1=50 \dimen2=\count1pt \divide\count1 by 8 \skip2=-10pt plus\count1fil minus\dimen2 \multiply\skip2 by-\count1 \divide\skip2 by \dimen2 \count6=\skip2 \skip1=.5\dimen2 plus\skip2 minus\count\count1fill \multiply\skip2 by\skip1 \advance\skip1 by-\skip2 \endtt \answer Set |\count1| to 50, then |\dimen2| to~$50\pt$, then |\count1| to~6, then |\skip2| to~$-10\pt$ plus~$6\,{\rm fil}$ minus~$50\pt$, then |\skip2| to~$60\pt$ plus~$-36\,{\rm fil}$ minus~$-300\pt$, then |\skip2| to~$1\,{\rm sp}$ minus~$-6\,{\rm sp}$, then |\count6| to~1, then |\skip1| to~$25\pt$ plus~$1\,{\rm sp}$ minus~$1\,{\rm fill}$, then |\skip2| to~$25\pt$ minus~$-150\pt$, then |\skip1| to~$0\pt$ plus~$1\,{\rm sp}$ minus~$1\,{\rm fill}$. \dangerexercise What is in |\skip5| after the following three commands have acted? \begintt \skip5=0pt plus 1pt \advance\skip5 by \skip4 \advance\skip5 by -\skip4 \endtt \answer If\/ |\skip4| has infinite stretchability, |\skip5| will be zero; otherwise it will be $0\pt$ plus~$1\pt$. \dangerexercise (For mathematicians.) Explain how to ^{round} |\dimen2| to the nearest multiple of\/ |\dimen3|, assuming that |\dimen3| is positive. \answer |\advance\dimen2 by\ifnum\dimen2<0 -\fi.5\dimen3|\parbreak |\divide\dimen2 by\dimen3 \multiply\dimen2 by\dimen3| \danger The registers obey \TeX's ^{group structure}. For example, changes to |\count3| inside |{...}| will not affect the value of |\count3| outside. Therefore \TeX\ effectively has more than 256 registers of each type. If you want the effect of a register command to transcend its group, you must say ^|\global| when you change the value. \dangerexercise What is in |\count1| after the following sequence of commands? \begintt \count1=5 {\count1=2 \global\advance\count1by\count1 \advance\count1by\count1} \endtt \answer |\count1| takes the values 5, then~2 (the old 5 is saved), then~4 (which is made global), then~8 (and 4~is saved); finally the value~4 is restored, and that is the answer. \ (For further remarks, see the discussion of\/ |\tracingrestores| in Chapter~27.) \danger The first ten |\count| registers, |\count0| through |\count9|, are reserved for a special purpose: \TeX\ displays these ten counts on your terminal whenever outputting a page, and it transmits them to the output file as an identification of that page. The counts are separated by decimal points on your terminal, with trailing `|.0|' patterns suppressed. Thus, for example, if |\count0=5| and |\count2=7| when a page is being shipped out to the ^|dvi| file, and if the other count registers are zero, \TeX\ will type `|[5.0.7]|'. Plain \TeX\ uses |\count0| for the page number, and it keeps |\count1| through~|\count9| equal to zero; that is why you see just `^|[1]|' when page~1 is being output. In more complex applications the page numbers can have further structure; ten counts are shipped out so that there will be plenty of identification. \danger It's usually desirable to have symbolic names for registers. \TeX\ provides a ^|\countdef| command (similar to |\chardef|, cf.~Chapter~8), which makes it easy to do this: You just say \begintt \countdef\chapno=28 \endtt and |\chapno| is henceforth an abbreviation for |\count28|. Similar commands ^|\dimendef|, ^|\skipdef|, and ^|\muskipdef| are available for the other types of numeric registers. After a control sequence has been defined by |\countdef|, it can be used in \TeX\ commands exactly as if it were an integer ^{parameter} like |\tolerance|. Similarly, |\dimendef| effectively creates a new dimension parameter, |\skipdef| effectively creates a new glue parameter, and |\muskipdef| effectively creates a new muglue parameter. \danger Besides the numerical registers, \TeX\ also has 256 box registers called ^|\box||0| to |\box255|. A~box register gets a value when you say ^|\setbox|\|=|^\; for example, `|\setbox3=\hbox{A}|' sets |\box3| to an hbox that contains the single letter~|A|. Several other examples of\/ |\setbox| have already appeared in Chapter~12. Chapter~10 points out that `|2\wd3|' is a \ that represents twice the width of\/ |\box3|; similarly, ^|\ht|\ and ^|\dp|\ can be used to refer to the height and depth of a given box register. ^^|\wd| \danger Box registers are local to groups just as arithmetic registers are. But there's a big difference between box registers and all the rest: When you use a |\box|, it loses its value. For example, the construction `|\raise2pt\box3|' in a horizontal list not only puts the contents of |\box3| into the list after raising it by~$2\pt$, it also makes |\box3| void. \TeX\ does this for efficiency, since it is desirable to avoid copying the contents of potentially large boxes. If you want to use a box register without wiping out its contents, just say `^|\copy|' instead of `|\box|'; for example, `|\raise2pt\copy3|'. \danger Another way to use a box register is to extract the inside of an hbox by saying `^|\unhbox|'. This annihilates the contents of the register, like `|\box|' does, and it also removes one level of boxing. For example, the commands \begintt \setbox3=\hbox{A} \setbox3=\hbox{\box3 B} \setbox4=\hbox{A} \setbox4=\hbox{\unhbox4 B} \endtt put |\hbox{\hbox{A}B}| into |\box3| and |\hbox{AB}| into |\box4|. Similarly, ^|\unvbox| unwraps a vbox. If you want to construct a large box by accretion (e.g., a ^{table of contents}), it is best to use |\unhbox| or |\unvbox| as in the |\setbox4| example; otherwise you use more of \TeX's memory space, and you might even obtain boxes inside boxes nested to such a deep level that hardware or software limits are exceeded. \danger The operations ^|\unhcopy| and ^|\unvcopy| are related to |\unhbox| and |\unvbox| as |\copy| is to |\box|. \ (But their names are admittedly peculiar.) \danger An unboxing operation ``unsets'' any glue that was set at the box's outer level. For example, consider the sequence of commands \begintt \setbox5=\hbox{A \hbox{B C}} \setbox6=\hbox to 1.05\wd5{\unhcopy5} \endtt This makes |\box6| five percent wider than |\box5|; the glue between |A| and |\hbox{B C}| stretches to make the difference, but the glue inside the inner hbox does not change. \danger A box register is either ``^{void}'' or it contains an hbox or a vbox. There is a difference between a void register and one that contains an empty box whose height, width, and depth are zero; for example, if\/ |\box3| is void, you can say |\unhbox3| or |\unvbox3| or |\unhcopy3| or |\unvcopy3|, but if\/ |\box3| is equal to |\hbox{}| you can say only |\unhbox3| or |\unhcopy3|. If you say `|\global\setbox3=|\', register |\box3| will become ``globally void'' when it is subsequently used or unboxed. \dangerexercise What is in register |\box5| after the following commands? \begintt \setbox5=\hbox{A} \setbox5=\hbox{\copy5\unhbox5\box5\unhcopy5} \endtt \answer |\hbox{\hbox{A}A}|. After `|\unhbox5|', |\box5| is void; |\unhcopy5| yields nothing. \dangerexercise And what's in |\box3| after ^^{grouping with box registers} `|{\global\setbox3=\hbox{A}\setbox3=\hbox{}}|'?\kern-1pt \answer |\hbox{A}|. But after `|{\global\setbox3=\hbox{A}\setbox3=\box3}|', |\box3| will be void. \danger If you are unsure about how \TeX\ operates on its registers, you can experiment online by using certain `|\show|' commands. For example, \begintt \showthe\count1 \showthe\dimen2 \showthe\skip3 \endtt will display the contents of\/ |\count1|, |\dimen2|, and |\skip3|; and `^|\showbox||4|' will display the contents of\/ |\box4|. ^^|\showthe| Box contents will appear only in the log file, unless you say `|\tracingonline=1|'. Plain \TeX\ provides a macro `^|\tracingall|' that turns on every possible mode of interaction, including ^|\tracingonline|. The author used these features to check the answers to several of the exercises above. \danger Large applications of \TeX\ make use of different sets of macros written by different groups of people. Chaos would reign if a register like |\count100|, say, were being used simultaneously for different purposes in different macros. Therefore plain \TeX\ provides an ^{allocation} facility; cooperation will replace confusion if each ^{macro writer} uses these conventions. The idea is to say, e.g., `^|\newcount|' when you want to dedicate a |\count| register to a special purpose. For example, the author designed a macro called `|\exercise|' to format the exercises in this manual, and one of the features of\/ |\exercise| is that it computes the number of the current exercise. The format macros in Appendix~E reserve a |\count| register for this purpose by saying \begintt \newcount\exno \endtt and then the command `|\exno=0|' is used at the beginning of each chapter. Similarly, `|\advance\exno by1|' is used whenever a new exercise comes along, and `|\the\exno|' is used to typeset the current exercise number. The |\newcount| operation assigns a unique count register to its argument |\exno|, and it defines |\exno| with a ^|\countdef| command. All of the other format macros are written without the knowledge of exactly which |\count| register actually corresponds to |\exno|. \danger Besides |\newcount|, plain \TeX\ provides ^|\newdimen|, ^|\newskip|, ^|\newmuskip|, and ^|\newbox|; there also are ^|\newtoks|, ^|\newread|, ^|\newwrite|, ^|\newfam|, and ^|\newinsert|, for features we haven't discussed yet. Appendices~B and~E contain several examples of the proper use of allocation. In the cases of |\newbox|, |\newread|, etc., the allocated number is defined by |\chardef|. For example, if the command `|\newbox\abstract|' is used to define a box register that will contain an abstract, and if the |\newbox| operation decides to allocate |\box45| for this purpose, then it defines the meaning of\/ |\abstract| by saying `|\chardef\abstract=45|'. \TeX\ allows ^|\chardef|'d quantities to be used as integers, so that you can say |\box\abstract| and |\copy\abstract|, etc. \ (There is no |\boxdef| command.) \newcount\notenumber \def\clearnotenumber{\notenumber=0\relax} \def\note{\advance\notenumber by1 \footnote{$^{\the\notenumber}$}} \clearnotenumber \dangerexercise Design a |\note| macro that produces footnotes numbered sequentially. For example,\note{First note.} it should produce the footnotes here\note{Second note.} if you type \begintt ... example,\note{First note.} it should produce the footnotes here\note{Second note.} if ... \endtt (Use |\newcount| to allocate a |\count| register for the footnotes.) \answer |\newcount\notenumber|\parbreak |\def\clearnotenumber{\notenumber=0\relax}|\parbreak |\def\note{\advance\notenumber by 1|\parbreak | \footnote{$^{\the\notenumber}$}}| \danger Sometimes, however, you want to use a register just for temporary storage, and you know that it won't conflict with anybody else's macros. Registers |\count255|, |\dimen255|, |\skip255|, and |\muskip255| are traditionally kept available for such purposes. Furthermore, plain \TeX\ reserves |\dimen0| to |\dimen9|, |\skip0| to |\skip9|, |\muskip0| to |\muskip9|, and |\box0| to |\box9| for ``scratchwork''; these registers are never allocated by the |\new...|\null\ operations. We have seen that |\count0| through |\count9| are special, and |\box255| also turns out to be special; so those registers should be avoided unless you know what you are doing. \ddanger Of course any register can be used for short-term purposes inside a group (including |\count0| to |\count9| and |\box255|, and including registers that have been allocated for other purposes), since register changes are local to ^{groups}. However, you should be sure that \TeX\ will not output any pages before the group has ended, because output routines might otherwise be invoked at unfortunate times. \TeX\ is ^^{output routines, when invoked}^^{page builder, when exercised} liable to invoke an output routine whenever it tries to move something from the list of recent contributions to the current page, because it might discover a page break with $c=\infty$ then. Here is a list of the times when that can happen: (a)~At the beginning or end of a paragraph, provided that this paragraph is being contributed to the main vertical list. (b)~At the beginning or end of a displayed equation within such a paragraph. (c)~After completing an |\halign| in vertical mode. (d)~After contributing a box or penalty or insertion to the main vertical list. (e)~After an |\output| routine has ended. \ddanger Now that we are armed with the knowledge of \TeX's flexible registers, we can plunge into the details of insertions. There are 255 classes of insertions, |\insert0| to |\insert254|, and they are tied to other registers of the same number. For example, |\insert100| is connected with |\count100|, |\dimen100|, |\skip100|, and |\box100|. Therefore plain \TeX\ provides an allocation function for insertions as it does for registers; Appendix~B includes the command \begintt \newinsert\footins \endtt ^^|\newinsert| which defines |\footins| as the number for footnote insertions. Other commands that deal with footnotes refer to |\count\footins|, |\dimen\footins|, and so on. The macros for floating topinserts are similarly prefaced by `|\newinsert\topins|', which defines |\topins| as the number of their class. Each class of insertions is independent, but \TeX\ preserves the order of insertions within a class. It turns out that |\footins| is class~254, and |\topins| is class~253, but the macros do not use such numbers directly. \def\n{\thinspace$n$} \ddanger For our purposes let's consider a particular class of insertions called class $n$; we will then be dealing with \TeX's primitive command ^^|\insert| \begindisplay |\insert|\n|{|\|}| \enddisplay which puts an insertion item into a horizontal or vertical list. For this class of insertions \begindisplay |\box|\n\ is where the material appears when a page is output;\cr |\count|\n\ is the magnification factor for page breaking;\cr |\dimen|\n\ is the maximum insertion size per page;\cr |\skip|\n\ is the extra space to allocate on a page.\cr \enddisplay For example, material inserted with |\insert100| will eventually appear in |\box100|. \ddanger Let the natural height plus depth of\/ |\insert|\n\ be $x$; then |\count|\n\ is 1000 times the factor by which $x$~affects the page goal. For example, plain \TeX\ sets |\count\footins=1000|, since there is a one-to-one relationship: A 10-point footnote effectively makes a page $10\pt$ shorter. But if we have an application where footnotes appear in double columns, a count value of 500 would be appropriate. One of the insertion classes in Appendix~E makes marginal notes for proofreading purposes; in that case the count value is zero. No actual magnification is done; |\count|\n\ is simply a number used for bookkeeping, when estimating the costs of various page breaks. \ddanger The first footnote on a page requires extra space, since we want to separate the footnotes from the text, and since we want to output a horizontal rule. Plain \TeX\ sets `|\skip\footins=|^|\bigskipamount|'; this means that a bigskip of extra space is assumed to be added by the output routine to any page that contains at least one insertion of class |\footins|. \ddanger Sometimes it is desirable to put a maximum limitation on the size of insertions; for example, people usually don't want an entire page to consist of footnotes. Plain \TeX\ sets |\dimen\footins=8in|; this means that |\box\footins| is not supposed to accumulate more than 8~inches of footnotes for any one page. \ddanger You might want to review the page-breaking algorithm explained at the beginning of this chapter, before reading further. On the other hand, maybe you don't really want to read the rest of this chapter at all, ever. \ddanger Here now is the algorithm that \TeX\ performs when an |\insert|\n\ is moved from the ``recent contributions'' to the ``current page.'' \ (Remember that such a move does not mean that the insertion will actually take place; the current page will be backed up later, to the breakpoint of least cost, and only the insertions preceding that breakpoint will actually be performed.) \ Let $g$ and $t$ be the current ^|\pagegoal| and ^|\pagetotal|; let $q$ be the ^|\insertpenalties| accumulated for the current page; and let $d$ and $z$ be the current ^|\pagedepth| and ^|\pageshrink|. \ (The value of~$d$ is at most ^|\maxdepth|; this value has not yet been incorporated into $t$.) \ Finally, let $x$ be the natural height plus depth of the |\insert|\n\ that we are moving to the current page; and let $f$ be the corresponding magnification factor, i.e., |\count|\n\ divided by 1000. \ninepoint \textindent{\bf Step 1.} If there is no previous |\insert|\n\ on the current page, decrease $g$ by $hf+w$, where $h$ is the current height plus depth of\/ |\box|\n, and where $w$ is the natural space component of |\skip|\n; also include the stretch and shrink components of |\skip|\n\ in the totals for the current page (in particular, this affects~$z$). \medbreak \textindent{\bf Step 2.} If a previous |\insert|\n\ on the current page has been split, add the parameter called ^|\floatingpenalty| to~$q$, and omit Steps 3 and~4. \medbreak \textindent{\bf Step 3.} Test if the current insertion will fit on the page without splitting. This means that it won't make the natural height-plus-depth of\/ |\box|\n\ surpass |\dimen|\n, when it is added to |\box|\n\ together with all previous |\insert|\n\ amounts on the current page; furthermore, it means that either $xf\le0$ or $t+d+xf-z\le g$. If both tests are passed, subtract $xf$ from~$g$ and omit Step~4. \medbreak \textindent{\bf Step 4.} (The current insertion will be split, at least tentatively; but the split will not actually take place if the least-cost page turns out to have occurred earlier than the present insertion.) \ First compute the largest amount~$v$ such that a height plus depth of $v$ will not make the total insertions into |\box|\n\ bigger than |\dimen|\n, and such that $t+d+vf\le g$. \ (Notice that $z$ is omitted from the latter formula, but the available shrinkability was considered in Step~3 when we tried to avoid splitting.) \ Then find the least-cost way to split the beginning of the vertical list of the insertion so as to obtain a box of height~$v$. \ (Use an algorithm just like page-breaking, but without the complexity of insertion; an additional `|\penalty-10000|' item is assumed to be present at the end of the vertical list, to ensure that a legal breakpoint exists.) \ Let $u$ be the natural height plus depth of that least-cost box, and let $r$ be the penalty associated with the optimum breakpoint. Decrease $g$ by~$uf$, and increase $q$ by~$r$. \ (If |\tracingpages||=1|, the log file should now get a cryptic message that says `|% split|\n\ |to| $v$|,|$u$ |p=|$r$'. For~example,^^{split insertion penalty} \begintt % split254 to 180.2,175.3 p=100 \endtt ^^{percent split} means that \TeX\ has tried to split an |\insert254| to height $180.2\pt$; the natural height-plus-depth of the best such split is $175.3\pt$, and the penalty for breaking there is~100.) \ddanger This algorithm is admittedly complicated, but no simpler mechanism seems to do nearly as much. Notice that penalties of $-10000$ inside insertions will make certain splits very attractive in Step~4, so the user can provide hints about where to break, in difficult situations. The algorithm provides a variety of different behaviors: Floating insertions can be accommodated as a special case of split insertions, by making each floating topinsert start with a small penalty, and by having zero as the associated |\floatingpenalty|; non-floating insertions like footnotes are accommodated by associating larger penalties with split insertions (see Appendix~B). \ddanger The splitting operation mentioned in Step 4 is also available as a primitive: `^|\vsplit|\ |to|\' produces a vbox obtained by splitting off a speci\-fied amount of material from a box register. For example, \begintt \setbox200=\vsplit100 to 50pt \endtt sets |\box200| to a vbox whose height is $50\pt$; it goes through the vertical list inside |\box100| (which should be a vbox) and finds the least-cost break assuming a goal height of~$50\pt$, considering badnesses and penalties just as in the case of page-breaking (but with $q=0$). The algorithm uses ^|\splitmaxdepth| instead of\/ ^|\maxdepth| to govern the maximum depth of boxes. Then it prunes the top of\/ |\box100| by removing everything up to and including any ^{discardable} items that immediately follow the optimum breakpoint; and it uses ^|\splittopskip| to insert new glue before the first box inside |\box100|, just as ^|\topskip| glue appears at the top of a page. However, if the optimum breakpoint occurs at the end of the vertical list inside |\box100|---a `|\penalty-10000|' item is assumed to be present there---or if all items after the optimum breakpoint are discarded, |\box100| will be void after the |\vsplit|. And if\/ |\box100| was void before the |\vsplit|, both |\box100| and |\box200| will be void afterwards. \ddanger \looseness=-1 You'd better not change |\box|\n, |\count|\n, |\dimen|\n, or |\skip|\n\ while \TeX\ is contributing insertions to the current page, since \TeX's algorithm assumes that those quantities are static. But you can change ^|\floatingpenalty|, |\splittopskip|, and |\splitmaxdepth|; \TeX\ will use the values that were current just inside the closing right brace of `|\insert|\n|{...}|' when it splits and floats insertions. For example, Appendix~B uses |\floatingpenalty=20000| in footnote insertions, to discourage footnotes that split before others can start, \vadjust{\break}but |\floatingpenalty=0| in floating topinserts. Appendix~B also uses special values of\/ |\splittopskip| and |\splitmaxdepth|, together with ^{struts}, so that split footnotes will be typeset with the same spacing as unsplit ones. \ddanger The |\footnote| macro puts an |\insert| into the horizontal list of a paragraph. After the paragraph has been broken into lines, this insertion will move out into the vertical list just after the line that contained it (see Chapter~14). Since there is no legal breakpoint between that box (i.e., that line) and the insertion, \TeX\ will put the insertion onto the page that contains the line that contains the insertion. \ddangerexercise Study the page-breaking algorithm carefully. Is it possible that a footnote might not appear on the same page as its reference? \answer Yes, in severe circumstances. (1)~Previous footnotes might have left no room for any more footnotes on the page. (2)~If |\vadjust{\eject}| occurs on the same line as a footnote, before that footnote, the reference will be forcibly detached. (3)~Other |\vadjust| commands on that line could also interpose breakpoints before the insertion. \ddanger \looseness=-1 When the best page break is finally chosen, \TeX\ removes everything after the chosen breakpoint from the bottom of the ``current page,'' and puts it all back at the top of the ``recent contributions.'' The chosen breakpoint itself is placed at the very top of the recent contributions. If it is a penalty item, the value of the penalty is recorded in ^|\outputpenalty| and the penalty in the contribution list is changed to $10000$; otherwise |\outputpenalty| is set to 10000. The insertions that remain on the current page are of three kinds: For each class~$n$ there are unsplit insertions, followed possibly by a single split insertion, followed possibly by others. If ^|\holdinginserts|$\null>0$, all insertions remain in place (so that they might be contributed again); otherwise they are all removed from the current page list as follows: The unsplit insertions are appended to |\box|\n, with no ^{interline glue} between them. \ (^{Struts} should be used, as in the |\vfootnote| macro of Appendix~B\null.) \ If a split insertion is present, it is effectively |\vsplit| to the size that was computed previously in Step~4; the top part is treated as an unsplit insertion, and the remainder (if any) is converted to an insertion as if it had not been split. This remainder, followed by any other floating insertions of the same class, is held over in a separate place. \ (They will show up on the ``current page'' if ^|\showlists| is used while an ^|\output| routine is active; the total number of such insertions appears in ^|\insertpenalties| during an |\output| routine.) %\ Finally, the remaining items before the best break on the current page are put together in a |\vbox| of height~$g$, where $g$ was the |\pagegoal| at the time of the break, using the saved value of\/ ^|\maxdepth|; this box becomes |\box255|. Now the user's ^|\output| ^^|\box255| routine enters \TeX's scanner (see Chapter~23); its duty is to assemble the final pages based on the contents of\/ |\box255| and any insertion boxes that it knows about. The output routine will probably unbox those boxes, so that their glue can be reset; the glue in insertion boxes usually cooperates nicely with the glue on the rest of the page, when it is given a chance. After the |\output| routine is finished, ^{held-over insertion} items are placed first on the list of recent contributions, followed by the vertical list constructed by |\output|, followed by the recent contributions beginning with the page break. \ (Deep breath.) \ You got that? \endchapter Since it is impossible to foresee how [footnotes] will happen to come out in the make-up, it is impracticable to number them from 1 up on each page. The best way is to number them consecutively throughout an article or by chapters in a book. \author UNIVERSITY OF ^{CHICAGO} PRESS, {\sl Manual of Style\/} (1910) % p102 \bigskip Don't use footnotes in your books, Don. \author JILL ^{KNUTH} (1962) \eject \beginchapter Chapter 16. Typing\\Math Formulas \TeX\ is designed to handle complex ^{mathematical expressions} in such a way that most of them are easy to input. The basic idea is that a complicated formula is composed of less complicated ^{formulas} put together in a simple way; the less complicated formulas are, in turn, made up of simple combinations of formulas that are even less complicated; and so on. Stating this another way, if you know how to type simple formulas and how to combine formulas into larger ones, you will be able to handle virtually any formula at all. So let's start with simple ones and work our way up. The simplest formula is a single letter, like `$x$', or a single number, like `2'. In order to put these into a \TeX\ text, you type `|$x$|' and `|$2$|', respectively. Notice that all mathematical formulas are enclosed in special math brackets; we are using |$| as the math bracket in this manual, in accord with the plain \TeX\ format defined in Appendix~B\null, because mathematics is supposedly expensive. When you type `|$x$|' the `$x$' comes out in italics, but when you type `|$2$|' the `$2$' comes out in roman type. In general, all characters on your keyboard have a special interpretation in math formulas, according to the normal conventions of mathematics printing: Letters now denote ^{italic} letters, while digits and punctuation denote ^{roman} digits and punctuation; a hyphen ({\tt-}) now denotes a ^{minus sign} ($-$), which is almost the same as an em-dash but not quite (see Chapter~2). The first |$| that you type puts you into ``^{math mode}'' and the second takes you out (see Chapter~13). So if you forget one |$| or type one |$| too many, \TeX\ will probably become thoroughly confused and you will probably get some sort of error message. ^^{dollarsign} Formulas that have been typeset by a printer who is unaccustomed to mathematics usually look quite strange to a mathematician, because a novice printer usually gets the spacing all wrong. In order to alleviate this problem, \TeX\ does most of its own spacing in math formulas; and it {\sl ignores\/} any ^{spaces} that you yourself put between |$|'s. For example, if you type `|$ x$|' and `|$ 2 $|', they will mean the same thing as `|$x$|' and `|$2$|'. You can type `\hbox{|$(x + y)/(x - y)$|}' or `|$(x+y) / (x-y)$|', but both will result in `$(x+y)/(x-y)$', a formula in which there is a bit of extra space surrounding the $+$ and~$-$ signs but none around the~/~sign. Thus, you do not have to memorize the complicated rules of math spacing, and you are free to use blank spaces in any way you like. Of course, spaces are still used in the normal way to mark the end of control sequences, as explained in Chapter~3. In most circumstances \TeX's spacing will be what a mathematician is accustomed to; but we will see in Chapter~18 that there are control sequences by which you can override \TeX's spacing rules if you want to. One of the things mathematicians like to do is make their formulas look like ^{Greek} to the uninitiated. In plain \TeX\ language you can type `|$$\alpha, \beta, \gamma, \delta;$$|' and you will get the first four Greek letters ^^|\alpha|^^|\beta|^^|\gamma|^^|\delta| $$\alpha,\beta,\gamma,\delta;$$ furthermore there are uppercase Greek letters like `$\Gamma$', which you can get by typing `|$\Gamma$|'. ^^|\Gamma| Don't feel intimidated if you aren't already familiar with Greek letters; they will be easy to learn if you need them. The only difficulty is that some symbols that look nearly the same must be carefully distinguished. For example, the Greek letters ^|\nu|~($\nu$) and ^|\kappa|~($\kappa$) should not be confused with the italic letters $v$ and~$x$; the Greek ^|\phi|~($\phi$) is different from the slashed zero called ^|\emptyset|~($\emptyset$). A~lowercase epsilon ($\epsilon$) is quite different from the symbol used to denote membership in a set ($\in$); type `|$\epsilon$|' for $\epsilon$ and `|$\in$|' for $\in$. ^^|\epsilon| ^^|\in| Some of the lowercase Greek letters have variant forms in plain \TeX's math italic fonts: `|$(\phi,\theta,\epsilon,\rho)$|' yields `$(\phi,\theta,\epsilon,\rho)$' while `|$(\varphi,\vartheta,\varepsilon,\varrho)$|' yields `$(\varphi,\vartheta,\varepsilon,\varrho)$'. ^^|\phi|^^|\theta|^^|\rho|^^|\varphi|^^|\vartheta|^^|\varrho|^^|\varepsilon| Besides Greek letters, there are a lot of ^{funny symbols} like `$\approx$' (which you get by typing `|$\approx$|') ^^|\approx|^^{special symbols for math} and `$\mapsto$' (which you ^^|\mapsto|^^{math symbols} get by typing `|$\mapsto$|'). A complete list of these control sequences and the characters they correspond to appears in Appendix~F\null. Such control sequences are allowed only in math mode, i.e., between |$|'s, because the corresponding symbols appear in the math fonts. \exercise What should you type to get the formula `$\gamma+\nu\in\Gamma$'\thinspace? \answer |$\gamma+\nu\in\Gamma$|. \exercise Look at Appendix F to discover the control sequences for `$\le$', `$\ge$', and~`$\ne$'. \ (These are probably the three most commonly used math symbols that are not present on your keyboard.) \ What does plain \TeX\ call them? \answer ^|\le|, ^|\ge|, and ^|\ne|. \ (These are short for ``less-or-equal,'' ``greater-or-equal,'' and ``not-equal.'') \ You can also use the names ^|\leq|, ^|\geq|, and ^|\neq|. \ (The fourth most common symbol is, perhaps, `$\infty$', which stands for ``^{infinity}'' and is called `^|\infty|'.) Now let's see how the more complex formulas get built up from simple ones. In the first place, you can get ^{superscripts} $\rm^{(up\,high)}$ and ^{subscripts} $\rm_{(down\,low)}$ ^^{indices, see subscripts} ^^{superiors, see superscripts} ^^{inferiors, see subscripts} by using `|^|' and `|_|', as shown in the following examples: \beginmathdemo \it Input&\it Output\cr \noalign{\vskip2pt} |$x^2$|&x^2\cr |$x_2$|&x_2\cr |$2^x$|&2^x\cr |$x^2y^2$|&x^2y^2\cr |$x ^ 2y ^ 2$|&x ^ 2y ^ 2\cr |$x_2y_2$|&x_2y_2\cr |$_2F_3$|&_2F_3\cr \endmathdemo Notice that |^| and |_| apply only to the next single character. If you want several things to be superscripted or subscripted, just enclose them in braces: \beginmathdemo |$x^{2y}$|&x^{2y}\cr |$2^{2^x}$|&2^{2^x}\cr |$2^{2^{2^x}}$|&2^{2^{2^x}}\cr |$y_{x_2}$|&y_{x_2}\cr |$y_{x^2}$|&y_{x^2}\cr \endmathdemo The braces in these examples have been used to specify ``^{subformulas},'' i.e., simpler parts of a larger formula. \TeX\ makes a box for each subformula, and treats that box as if it were a single symbol. Braces also serve their usual purpose of grouping, as discussed in Chapter~5. It is illegal to type `|x^y^z|' or `|x_y_z|'; \TeX\ will complain of a ``double superscript'' or ``double subscript.'' You must type `|x^{y^z}|' or `|x^{yz}|' or `|x_{y_z}|' or `|x_{yz}|' in order to make your intention clear. A superscript or subscript following a character applies to that character only; but when following a subformula it applies to that whole subformula, and it will be raised or lowered accordingly. For example, \beginmathdemo |$((x^2)^3)^4$|&((x^2)^3)^4\cr |${({(x^2)}^3)}^4$|&{({(x^2)}^3)}^4\cr \endmathdemo In the first formula the `|^3|' and `|^4|' are superscripts on the ^{right parentheses}, i.e., on the `|)|' characters that immediately precede them, but in the second formula they are superscripts on the subformulas that are enclosed in braces. The first alternative is preferable, because it is much easier to type and it is just as easy to read. \danger A subscript or superscript following nothing (as in the `|_2F_3|' example on the preceding page, where the `|_2|' follows nothing) is taken to mean a subscript or superscript of an empty subformula. Such notations are (fortunately) rare in mathematics; but if you do encounter them it is better to make your intention clear by showing the empty subformula explicitly with braces. In other words, the best way to get `${}_2F_3$' in a formula is to type `|{}_2F_3|' or `|{_2}F_3|' or `|{_2F_3}|'. \dangerexercise What difference, if any, is there between the output of `|$x + _2F_3$|' and the output of `|$x + {}_2F_3$|'\thinspace? \answer In the former, the `|_2|' applies to the plus sign ($x + _2F_3$); but in the latter, it applies to an empty subformula ($x + {}_2F_3$). \dangerexercise Describe the differences between the outputs of `|${x^y}^z$|' and `|$x^{y^z}$|'. \answer The results are `${x^y}^z$' and `$x^{y^z}$'; the $z$ in the first alternative is the same size as the $y$, but in the second it is smaller. Furthermore, the $y$ and $z$ in the first case aren't quite at the same height. \ (Good typists never even think of the first construction, because mathematicians never want it.) You can have simultaneous subscripts and superscripts, and you can specify them in any order: \beginmathdemo |$x^2_3$|&x^2_3\cr |$x_3^2$|&x_3^2\cr |$x^{31415}_{92}+\pi$|&x^{31415}_{92}+\pi\cr \noalign{\smallskip} |$x_{y^a_b}^{z_c^d}$|&x_{y^a_b}^{z_c^d}\cr \endmathdemo Notice that simultaneous su$\rm_b^{per\kern-1pt}$scripts are positioned over each other. However, a subscript will be ``tucked in'' slightly when it follows certain letters; for example, `|$P_2^2$|' produces `$P_2^2$'. If for some reason you want the left edges of both subscript and superscript to be aligned, you can fool \TeX\ by inserting a null subformula: `|$P{}_2^2$|' produces `$P{}_2^2$'. The control sequence ^|\prime| stands for the symbol `$\prime$', which is used mostly in superscripts. In fact, `$\prime$' is so big as it stands that you would never want to use it except in a subscript or superscript, where it occurs in a smaller size. Here are some typical examples: \beginmathdemo \it Input&\it Output\cr \noalign{\vskip2pt} |$y_1^\prime$|&y_1^\prime\cr |$y_2^{\prime\prime}$|&y_2^{\prime\prime}\cr |$y_3^{\prime\prime\prime}$|&y_3^{\prime\prime\prime}\cr \endmathdemo Since single and double primes occur rather frequently, plain \TeX\ provides a convenient abbreviation: You can simply type |'| instead of |^\prime|, and |''| instead of |^{\prime\prime}|, and so on. \beginmathdemo |$f'[g(x)]g'(x)$|&f'[g(x)]g'(x)\cr |$y_1'+y_2''$|&y_1'+y_2''\cr |$y'_1+y''_2$|&y'_1+y''_2\cr |$y'''_3+g'^2$|&y'''_3+g'{}^2\cr \endmathdemo \dangerexercise Why do you think \TeX\ treats |\prime| as a large symbol that appears only in superscripts, instead of making it a smaller symbol that has already been shifted up into the superscript position? \answer The second alternative doesn't work properly when there's a subscript at the same time as a prime. Furthermore, some mathematicians use |\prime| also in the subscript position; they write, for example, $F'(w,z)=\partial F(w,z)/\partial z$ and $F_\prime(w,z)=\partial F(w,z)/ \partial w$. \dangerexercise Mathematicians sometimes use ``^{tensor notation}'' in which subscripts and superscripts are staggered, as in `$R_i{}^{jk}{}_l$'. Explain how to achieve such an effect. \answer |$R_i{}^{jk}{}_l$|. Another way to get complex formulas from simple ones is to use the control sequences ^|\sqrt|, ^|\underline|, or ^|\overline|. ^^{surds, see sqrt} ^^{vinculum, see overline} Like |^| and |_|, these operations apply to the character or subformula that follows them: \beginmathdemo |$\sqrt2$|&\sqrt2\cr |$\sqrt{x+2}$|&\sqrt{x+2}\cr |$\underline4$|&\underline4\cr |$\overline{x+y}$|&\overline{x+y}\cr |$\overline x+\overline y$|&\overline x+\overline y\cr |$x^{\underline n}$|&x^{\underline n}\cr |$x^{\overline{m+n}}$|&x^{\overline{m+n}}\cr |$\sqrt{x^3+\sqrt\alpha}$|&\sqrt{x^3+\sqrt\alpha}\cr \endmathdemo You can also get cube roots `$\root3\of{\phantom{h}}$' and similar things by using ^|\root|: \beginmathdemo |$\root 3 \of 2$|&\root 3 \of 2\cr |$\root n \of {x^n+y^n}$|&\root n \of {x^n+y^n}\cr |$\root n+1 \of a$|&\root n+1 \of a\cr \endmathdemo \danger The |\sqrt| and |\underline| and |\overline| operations are able to place lines above or below subformulas of any size or shape; the bar lines change their size and position, so that they are long enough to cover the subformula, and high enough or low enough not to bump into it. For example, consider `|\overline|~|l|' ($\,\overline l\,$) versus `|\overline|~|m|' ($\,\overline m\,$): The first has a shorter bar line, and this line has been raised higher than the bar in the second. Similarly, the bar in `|\underline|~|y|' ($\,\underline y\,$) is lower than the bar in `|\underline|~|x|' ($\,\underline x\,$); and square root signs appear in a variety of positions based on the height and depth of what is being |\sqrt|'d: $\sqrt a + \sqrt d + \sqrt y$. \TeX\ knows the height, depth, and width of every letter and every subformula, because it considers them to be boxes, as explained in Chapter~11. If you have a formula in which there is only one |\sqrt|, or only one |\overline| or |\underline|, the normal positioning rules work fine; but sometimes you want to have uniformity between different members of a complex formula. For example, you might want to typeset `$\sqrt{\mathstrut a}+\sqrt{\mathstrut d}+\sqrt{\mathstrut y}$', putting all square roots in the same vertical position. There's an easy way to do this, using the control sequence ^|\mathstrut| as follows: \begintt $\sqrt{\mathstrut a}+\sqrt{\mathstrut d}+\sqrt{\mathstrut y}$. \endtt A |\mathstrut| is an invisible box whose width is zero; its height and depth are the height and depth of a parenthesis `('. Therefore subformulas that contain |\mathstrut| will always have the same height and depth, unless they involve more complicated constructions like subscripts and superscripts. Chapter~18 discusses more powerful operations called ^|\smash| and ^|\phantom| by which you can obtain complete control over the positioning of roots and similar signs. \exercise Test your understanding of what you have read so far in this chapter by explaining what should be typed to get the following formulas. \ (Be sure to check your answer with Appendix~A to confirm that you're right.) $$\hbox to\hsize{\indent $\displaystyle 10^{10}\hfil 2^{n+1}\hfil (n+1)^2\hfil \sqrt{1-x^2}\hfil \overline{w+\overline z}\hfil p_1^{e_1}\hfil a_{b_{c_{d_e}}}\hfil \root3\of{h''_n(\alpha x)}\hfil$}$$ \answer |10^{10}|; \stretch|2^{n+1}|; \stretch|(n+1)^2|; \stretch|\sqrt{1-x^2}|; \stretch|\overline{w+\overline z}|; \stretch|p_1^{e_1}|; \stretch |a_{b_{c_{d_e}}}|; \stretch|\root3\of{h''_n(\alpha x)}|. \ (Of course, you should enclose these formulas in dollar signs so that \TeX\ will process them in math mode. Superscripts and subscripts can be given in either order; for example, |h''_n| and |h_n''| both work the same. You should not leave out any of the braces shown here; for example, `|$10^10$|' would yield `$10^10$'. But it doesn't hurt to insert additional braces around letters or numbers, as in `|({n}+{1})^{2}|'. The indicated blank spaces are necessary unless you use extra braces; otherwise \TeX\ will complain about undefined control sequences |\overlinez| and |\alphax|.) \exercise What mistake did B. C. ^{Dull} discover after he typed the following? \begintt If$ x = y$, then $x$ is equal to $y.$ \endtt \answer He got `If$ x = y\ldots$' because he forgot to leave a space after `|If|'; ^{spaces} disappear between dollar signs. He should also have ended the sentence with `|$y$.|'; punctuation that belongs to a sentence should not be included in a formula, as we will see in Chapter~18. \ (But you aren't expected to know that yet.) \exercise Explain how to type the following sentence: \begindisplay Deleting an element from an $n$-tuple leaves an $(n-1)$-tuple. \enddisplay \answer |Deleting an element from an $n$-tuple leaves an $(n-1)$-tuple.| \exercise List all the italic letters that descend below the baseline. \ (These are the letters for which |\underline| will lower its bar line.) \answer $Q,f,g,j,p,q,y$. \ (The analogous ^{Greek} letters are ^^{italic letters with descenders} ^^{descenders} $\beta,\gamma,\zeta,\eta,\mu,\xi,\rho,\phi,\varphi,\chi,\psi$.) We have discussed the fact that the characters you type have special meanings in math mode, but the examples so far are incomplete; they don't reveal all the power that is at your fingertips just after you press the `|$|' key. It's time now to go back to basics: Let us make a systematic survey of what each character does, when it is used in a formula. The 52 ^{letters} (|A| to |Z| and |a| to |z|) denote italic symbols ($A$~to~$Z$ and $a$~to~$z$), which a mathematician would call ``^{variables}.'' \TeX\ just calls them ``^{ordinary symbols},'' because they make up the bulk of math formulas. There are two variants of lowercase L in plain \TeX, namely `$l$' (which you get by simply typing `|l|') and `$\ell$' (which you get by typing `^|\ell|'). Although mathematicians commonly write something that looks like `$\ell$' in their manuscripts, they do so only to distinguish it from the numeral~`1'. This distinguishability problem is not present in printed mathematics, since an italic `$l$' is quite different from a~`1'; therefore it is traditional to use `$l$' unless `$\ell$' has been specifically requested. Plain \TeX\ also treats the 18 characters \begintt 0 1 2 3 4 5 6 7 8 9 ! ? . || / ` @ " \endtt ^^{digits}^^{numerals} as ordinary symbols; i.e., it doesn't insert any extra space when these symbols occur next to each other or next to letters. Unlike the letters, these 18 characters remain in roman type when they appear in formulas. There's nothing special for you to remember about them, except that the ^{vertical line} `\|' has special uses that we shall discuss later. Furthermore, you should be careful to distinguish between `oh' and `zero': The italic letter~$O$ is almost never used in formulas unless it appears just before a left parenthesis, as in `$O(n)$'; ^^{big-$O$ notation} and the numeral~$0$ is almost never used just before a left parenthesis unless it is preceded by another digit, as in `$10(n-1)$'. Watch for left parentheses and you'll be $0K$. \ (Lowercase o's also tend to appear only before left parentheses; type `|x_0|' instead of `|x_o|', since the formula `$x_0$' is generally more correct than `$x_o$'.) The three characters |+|, |-|, and |*| are called ``^{binary operations},'' because they operate on two parts of a formula. For example, |+|~is a ^{plus sign}, which is used for the sum of two numbers; |-|~is a ^{minus sign}. The ^{asterisk}~(|*|) is rarer ^^{star, see asterisk} in mathematics, but it also behaves as a binary operation. Here are some examples of how \TeX\ typesets binary operations when they appear next to ordinary symbols: \beginmathdemo \it Input&\it Output\cr \noalign{\vskip2pt} |$x+y-z$|&x+y-z\cr |$x+y*z$|&x+y*z\cr |$x*y/z$|&x*y/z\cr \endmathdemo Notice that |-| and |*| produce quite different math symbols from what you get in normal text: The ^{hyphen}~(-) becomes a minus sign~($-$), and the raised asterisk~(*) drops down to a lower level~($*$). \danger \TeX\ does not treat |/| as a binary operation, even though a ^{slash} stands for division (which qualifies as a binary operation on mathematical grounds). The reason is that printers traditionally put extra space around the symbols $+$, $-$, and~$*$, but not around~$/$. If \TeX\ were to typeset |/| as a binary operation, the formula `|$1/2$|' would come out `$1\mathbin/2$', which is wrong; so \TeX\ considers |/| to be an ordinary symbol. \danger Appendix F lists many more binary operations, for which you type control sequences instead of single characters. Here are some examples: \beginmathdemo |$x\times y\cdot z$|&x\times y\cdot z\cr |$x\circ y\bullet z$|&x\circ y\bullet z\cr |$x\cup y\cap z$|&x\cup y\cap z\cr |$x\sqcup y\sqcap z$|&x\sqcup y\sqcap z\cr |$x\vee y\wedge z$|&x\vee y\wedge z\cr |$x\pm y\mp z$|&x\pm y\mp z\cr \endmathdemo It is important to distinguish $\times$ (^|\times|) from $X$ (|X|) and from $x$ (|x|); to distinguish $\cup$ (^|\cup|) from $U$ (|U|) and from $u$ (|u|); to distinguish $\vee$ (^|\vee|) from $V$ (|V|) and from $v$ (|v|); to distinguish $\circ$ (^|\circ|) from $O$ (|O|) and from $o$ (|o|). ^^|\cdot|^^|\bullet|^^|\cap|^^|\sqcup|^^|\sqcap|^^|\wedge| ^^{cross, see dagger, times} ^^|\pm|^^|\mp| The symbols `$\lor$' and `$\land$' can also be called ^|\lor| and ^|\land|, since they frequently stand for binary operations that are called ``^{logical~or}'' and ``^{logical~and}.'' \danger Incidentally, binary operations are treated as ordinary symbols if they don't occur between two quantities that they can operate on. For example, no extra space is inserted next to the $+$, $-$, and~$*$ in cases like the following: \beginmathdemo |$x=+1$|&x=+1\cr |$3.142-$|&3.142-\cr |$(D*)$|&(D*)\cr \endmathdemo Consider also the following examples, which show that binary operations can be used as ordinary symbols in superscripts and subscripts: \beginmathdemo |$K_n^+,K_n^-$|&K_n^+,K_n^-\cr |$z^*_{ij}$|&z^*_{ij}\cr |$g^\circ \mapsto g^\bullet$|&g^\circ \mapsto g^\bullet\cr |$f^*(x) \cap f_*(y)$|&f^*(x) \cap f_*(y)\cr \endmathdemo \dangerexercise How would you obtain the formulas `$z^{*2}$' and `$h_*'(z)$'\thinspace? \answer |$z^{*2}$| and |$h_*'(z)$|. Plain \TeX\ treats the four characters |=|, |<|, |>|, and |:|\ as ``^{relations}'' because they express a relationship between two quantities. For example, `${xz$|&x=y>z\cr |$x:=y$|&x:=y\cr |$x\le y\ne z$|&x\le y\ne z\cr |$x\sim y\simeq z$|&x\sim y\simeq z\cr |$x\equiv y\not\equiv z$|&x\equiv y\not\equiv z\cr |$x\subset y\subseteq z$|&x\subset y\subseteq z\cr \endmathdemo ^^|\le|^^|\ne|^^|\simeq|^^{colon}^^{equals}^^{lessthan}^^{greaterthan} ^^{colonequals}^^|\equiv|^^|\not|^^|\subset|^^|\subseteq|^^|\sim| ^^{hooks, see subset, supset} ^^{wiggle, see sim} (The last several examples show some of the many other relational symbols that plain \TeX\ makes available via control sequences; see Appendix~F.) The two characters `|,|' (^{comma}) and `|;|' (^{semicolon}) are treated as ^{punctuation marks in formulas}; this means that \TeX\ puts a little extra space after them, but not before them. \beginmathdemo |$f(x,y;z)$|&f(x,y;z)\cr \endmathdemo It isn't customary to put extra space after a `|.|'\ (^{period}) in math formulas, so \TeX\ treats a period as an ordinary symbol. If you want the `|:|'\ character to be treated as a punctuation mark instead of as a relation, just call it ^|\colon|: \beginmathdemo |$f:A\to B$|&f:A\to B\cr |$f\colon A\to B$|&f\colon A\to B\cr \endmathdemo ^^|\to| If you want to use a comma as an ordinary symbol (e.g., when it appears in a large number), just put it in braces; \TeX\ treats anything in braces as an ordinary symbol. For instance, \beginmathdemo |$12,345x$|&12,345x\qquad\rm(wrong)\cr |$12{,}345x$|&12{,}345x\qquad\,\rm(right)\cr \endmathdemo \dangerexercise What's an easy way to get a raised dot in a decimal constant (e.g., `$3{\cdot}1416$')? \answer |$3{\cdot}1416$|. \ (One of the earlier examples in this chapter showed that ^|\cdot| is a binary operation; putting it in braces makes it act like an ordinary symbol.)\par If you have lots of constants like this, for example in a table, there's a way to make ordinary periods act like |\cdot| symbols: Just define ^|\mathcode||`.| to be |"0201|, assuming that the fonts of plain \TeX\ are being used. However, this could be dangerous, since ordinary periods are used frequently in displayed equations; the |\mathcode| change should be confined to places where every period is to be a |\cdot|. So far we have considered letters, other ordinary symbols, binary operations, relations, and punctuation marks; hence we have covered almost every key on the typewriter. There are just a few more: The characters `|(|' ^^{lparen} and `|[|' ^^{lbracket} are called ``^{openings},'' while `|)|' ^^{rparen} and ^^{fences, see opening, closing, delimiters} `|]|' ^^{rbracket} are called ``^{closings}''; these act pretty much like ordinary symbols, but they help \TeX\ to decide when a binary operation is not really being used in a binary way. Then there is the character~|'|, which we know is used as an abbreviation for |\prime| superscripts. Finally, we know that plain \TeX\ reserves the other ten characters: \begintt \ $ % # & ~ { } _ ^ \endtt These are not usable for symbols in math mode unless their ^|\catcode| values are changed (see Chapter~7). Although |{| and |}| specify grouping, the control sequences `|\{|' and `|\}|' ^^|\lbrace|^^|\rbrace| can be used to get `$\{$' as an opening and `$\}$' as a closing. \ddanger All of these math mode interpretations are easily changeable, since each character has a ^|\mathcode|, as explained in Chapter~17; none of the conventions are permanently built into \TeX\null. However, most of them are so standard that it is usually unwise to make many changes, except perhaps in the interpretations of |`|, |"|, and |@|. ^^{at sign} ^^{leftquote} ^^{doublequote} The special characters |^| and |_| that designate superscripts ^^{circumflex}^^{underbar} and subscripts should not be used except in formulas. Similarly, the names of math symbols like |\alpha| and |\approx|, and the control sequences for math operations like |\overline|, must not invade ordinary text. \TeX\ uses these facts to detect ^{missing dollar signs} in your input, before such mistakes cause too much trouble. For example, suppose you were to type \begintt The smallest $n such that $2^n>1000$ is~10. \endtt \TeX\ doesn't know that you forgot a `|$|' after the first `|n|', because it doesn't understand English; so it finds a ``formula'' between the first two |$| signs: \begindisplay The smallest $n such that $ \enddisplay after which it thinks that `|2|' is part of the text. But then the |^| reveals an inconsistency; \TeX\ will automatically insert a~|$| before the~|^|, and you will get an error message. In this way the computer has gotten back into synch, and the rest of the document can be typeset as if nothing had happened. \danger Conversely, a blank line or ^|\par| is not permitted in math mode. This gives \TeX\ another way to recover from a missing~|$|; such errors will be confined to the paragraph in which they occur. \danger If for some reason you cannot use |^| and |_| for superscripts and subscripts, because you have an unusual keyboard or because you need |^| for French accents or something, plain \TeX\ lets you type ^|\sp| and ^|\sb| instead. For example, `|$x\sp2$|' is another way to get `$x\sp2$'. On the other hand, some people are lucky enough to have keyboards that contain additional symbols besides those of standard ASCII. ^^{character set} When such symbols are available, \TeX\ can be set up to make math typing a bit more pleasant. For example, at the author's installation there are keys labeled \up\ and~\dn\ that produce visible symbols (these make superscripts and subscripts look much nicer on the screen); there are keys for the relations {\tentex\char'34}, {\tentex\char'35}, and {\tentex\char'32} (these save time); and there are about two dozen more keys that occasionally come in handy. \ (See Appendix~C.) ^^{uparrow}^^{downarrow}^^{leq}^^{geq}^^{neq} \danger Mathematicians are fond of using ^{accents} over letters, because this is often an effective way to indicate relationships between mathematical objects, and because it greatly extends the number of available symbols without increasing the number of necessary fonts. Chapter~9 discusses the use of accents in ordinary text, but mathematical accents are somewhat different, because spacing is not the same; \TeX\ uses special conventions for accents in formulas, so that the two sorts of accents will not be confused with each other. The following math accents are provided by plain~\TeX: \beginmathdemo |$\hat a$|&\hat a\cr |$\check a$|&\check a\cr |$\tilde a$|&\tilde a\cr |$\acute a$|&\acute a\cr |$\grave a$|&\grave a\cr |$\dot a$|&\dot a\cr |$\ddot a$|&\ddot a\cr |$\breve a$|&\breve a\cr |$\bar a$|&\bar a\cr |$\vec a$|&\vec a\cr \endmathdemo ^^|\hat|^^|\check|^^|\tilde|^^|\acute|^^|\grave| ^^|\dot|^^|\ddot|^^|\breve|^^|\bar|^^|\vec| The first nine of these are called |\^|, |\v|, |\~|, |\'|, |\`|, |\.|, |\"|, |\u|, and |\=|, respectively, when they appear in text; |\vec| is an accent that appears only in formulas. \TeX\ will complain if you try to use |\^| or |\v|, etc., in formulas, or if you try to use |\hat| or |\check|, etc., in ordinary text. \danger It's usually a good idea to define special control sequences for accented letters that you need frequently. For example, you can put ^^|\def| \begintt \def\Ahat{{\hat A}} \def\chat{{\hat c}} \def\scheck{{\check s}} \def\xtilde{{\tilde x}} \def\zbar{{\bar z}} \endtt at the beginning of a manuscript that uses the symbols $\hat A$, $\hat c$, $\check s$, $\tilde x$, and $\bar z$ more than, say, five times. This saves you a lot of keystrokes, and it makes the manuscript easier to read. Chapter~20 explains how to define control sequences. \danger When the letters $i$ and $j$ are accented in math formulas, ^{dotless} symbols $\imath$ and $\jmath$ should be used under the accents. These symbols are called ^|\imath| and ^|\jmath| in plain \TeX. Thus, for example, a paper that uses `$\hat\imath$' and `$\hat\jmath$' ought to begin with the following definitions: \begintt \def\ihat{{\hat\imath}} \def\jhat{{\hat\jmath}} \endtt \def\Ahat{{\hat A}} \danger You can put ^{accents on top of accents}, making symbols like $\skew6\hat\Ahat$ that might cause a mathematician to squeal with ecstasy. However, it takes a bit of finesse to get the upper accent into a position that looks right, because the designer of a font for mathematics usually tells \TeX\ to position math accents in special ways for special letters. Plain \TeX\ provides a control sequence called ^|\skew| that makes it fairly easy to shift superaccents into their proper place. For example, `|\skew6\hat\Ahat|' was used to produce the symbol above. The number `|6|' in this example was chosen by trial and error; `|5|'~seems to put the upper accent a bit too far left, while `|7|'~makes it a bit too far right, at least in the author's opinion. The idea is to fiddle with the amount of skew until you find what pleases you best. \danger It's possible, in fact, to put math accents on any subformula, not just on single characters or accented characters. But there's usually not much point in doing so, because \TeX\ just centers the accent over the whole subformula. For example, `|$\hat{I+M}$|' yields `$\hat{I+M}$'. In particular, a |\bar| accent always stays the same size; it's not like ^|\overline|, which grows with the formula under it. Some people prefer the longer line from |\overline| even when it applies to only a single letter; for example, `|$\bar z+\overline z$|' produces `$\bar z+\overline z$', and you can take your pick when you define |\zbar|. However, plain \TeX\ does provide two accents that grow; they are called ^|\widehat| and ^|\widetilde|: \beginmathdemo |$\widehat x,\widetilde x$|&\tenmath\widehat x,\widetilde x\cr |$\widehat{xy},\widetilde{xy}$|&\tenmath\widehat{xy},\widetilde{xy}\cr |$\widehat{xyz},\widetilde{xyz}$|&\tenmath\widehat{xyz},\widetilde{xyz}\cr \endmathdemo The third example here shows the maximum size available. \def\ghat{{\hat g}} \exercise This has been another long chapter; but cheer up, you have learned a lot! Prove it by explaining what to type in order to get the formulas $e^{-x^2}$, $D\sim p^\alpha M+l$, and $\ghat\in(H^{\pi_1^{-1}})'$. \ (In the last example, assume that a control sequence |\ghat| has already been defined, so that |\ghat| produces the accented letter $\ghat$.) \answer |$e^{-x^2}$|, |$D\sim p^\alpha M+l$|, and |$\ghat\in(H^{\pi_1^{-1}})'$|. \ (If you are reading the dangerous bend sections, you know that the recommended way to define |\ghat| is `|\def\ghat{{\hat g}}|'.) \endchapter Producing ^{Greek} letters is as easy as $\pi$. You just type |... as easy as $\pi$.| \author LESLIE ^{LAMPORT}, {\sl The ^{L\kern-.2em\raise.6ex\hbox{a}% \kern-.1em\TeX} Document Preparation System\/} (1983) % Note: the final manual has a slightly different wording on p43. % It's now called "LaTeX: A Document Preparation System" (1986) % But I decided to cite the original, partly because I have % no smallcaps sans-serif `A' to match the new LaTeX logo! \bigskip \TeX\ has no regard for the glories of the Greek tongue---\/ as far as it is concerned, Greek letters are just additional weird symbols, and they are allowed\/ {\rm only} in math mode. In a pinch you can get the output $\tau\epsilon\chi$ by typing % |$\tau\epsilon\chi$|, but if you're actually setting Greek text, you will be using a different version of \TeX, designed for a keyboard with Greek letters on it, and you shouldn't even be reading this manual, which is undoubtedly all English to you. \author MICHAEL ^{SPIVAK}, {\sl The Joy of \TeX\/} (1982) \eject \beginchapter Chapter 17. More about Math Another thing mathematicians like to do is make fractions---and they like to build symbols up on top of each other in a variety of different ways: \begindisplay $\displaystyle {1\over2}\qquad{\rm and}\qquad{n+1\over3}\qquad{\rm and}\qquad {n+1\choose3}\qquad{\rm and}\qquad\sum_{n=1}^3 Z_n^2\,.$ \enddisplay You can get these four formulas as displayed equations by typing `|$$1\over2$$|' and `|$$n+1\over3$$|' and `|$$n+1\choose3$$|' and `|$$\sum_{n=1}^3 Z_n^2$$|'; we shall study the simple rules for such constructions in this chapter. ^^|\sum|^^|\choose| First let's look at ^{fractions}, which use the `^|\over|' notation. The control sequence |\over| applies to everything in the formula unless you use braces to enclose it in a specific subformula; in the latter ^^{stacked fractions, see over} case, |\over| applies to everything in that subformula. \begindisplaymathdemo \it Input&\it Output\cr \noalign{\vskip-3pt} |$$x+y^2\over k+1$$|&x+y^2\over k+1\cr \noalign{\vskip2pt} |$${x+y^2\over k}+1$$|&{x+y^2\over k}+1\cr \noalign{\vskip-1pt} |$$x+{y^2\over k}+1$$|&x+{y^2\over k}+1\cr \noalign{\vskip-1pt} |$$x+{y^2\over k+1}$$|&x+{y^2\over k+1}\cr \noalign{\vskip-3pt} |$$x+y^{2\over k+1}$$|&x+y^{2\over k+1}\cr \endmathdemo You aren't allowed to use |\over| twice in the same subformula; instead of typing something like `|a \over b \over 2|', you must specify what goes over what: \begindisplaymathdemo \noalign{\vskip3pt} |$${a\over b}\over 2$$|&{a\over b}\over 2\cr |$$a\over{b\over 2}$$|&a\over{b\over 2}\cr \endmathdemo Unfortunately, both of these alternatives look pretty awful. Mathematicians tend to ``overuse'' |\over| when they first begin to typeset their own work on a system like \TeX. A good typist or copy editor will convert fractions to a ``^{slashed form},'' whenever a built-up construction would be too small or too crowded. For example, the last two cases should be treated as follows: \begindisplaymathdemo \noalign{\vskip3pt} |$$a/b \over 2$$|&a/b \over 2\cr |$$a \over b/2$$|&a \over b/2\cr \endmathdemo Conversion to slashed form takes a little bit of mathematical knowhow, since ^{parentheses} sometimes need to be inserted in order to preserve the meaning of the formula. Besides substituting `|/|' for~`|\over|', the two parts of the fraction should be put in parentheses unless they are single symbols; for example, $a\over b$~becomes simply~$a/b$, but $a+1\over b$ becomes $(a+1)/b$, and $a+1\over b+1$ becomes ${(a+1)/(b+1)}$. Furthermore, the entire fraction should generally be enclosed in parentheses if it appears next to something else; for example, ${a\over b}x$ becomes $(a/b)x$. If you are a typist without mathematical training, it's best to ask the author of the manuscript for help, in doubtful cases; you might also tactfully suggest that unsightly fractions be avoided altogether in future manuscripts. \exercise What's a better way to render the formula $x+y^{2\over k+1}$? \answer $x+y^{2/(k+1)}$\quad(|$x+y^{2/(k+1)}$|). \exercise Convert `${a+1\over b+1}x$' to slashed form. \answer $((a+1)/(b+1))x$\quad(|$((a+1)/(b+1))x$|). \exercise What surprise did B. L. ^{User} get when he typed `|$$x = (y^2\over k+1)$$|'\thinspace? \answer He got the displayed formula$$x=(y^2\over k+1)$$ because he forgot that an unconfined |\over| applies to everything. \ (He should probably have typed `|$$x=\left(y^2\over k+1\right)$$|', using ideas that will be presented later in this chapter; this not only makes the parentheses larger, it keeps the `$x=$' out of the fraction, because |\left| and |\right| introduce subformulas.) \def\cents{\hbox{\rm\rlap/c}} \exercise How can you make `$7{1\over2}\cents$'? \ (Assume that the control sequence |\cents| yields~`$\cents$'.)^^{money}^^{cents} \answer `|$7{1\over2}\cents$|' or `|7$1\over2$\cents|'. \ (Incidentally, the definition used here was |\def\cents{\hbox{\rm\rlap/c}}|.) ^^|\rlap|^^|\cents| The examples above show that letters and other symbols sometimes get smaller when they appear in fractions, just as they get smaller when they are used as exponents. It's about time that we studied \TeX's method for choosing the sizes of things. \TeX\ actually has eight different ^{styles} in which it can treat formulas, namely $$\halign{\indent#\hfil\quad&#\hfil\cr display style&(for formulas displayed on lines by themselves)\cr text style&(for formulas embedded in the text)\cr script style&(for formulas used as superscripts or subscripts)\cr scriptscript style&(for second-order superscripts or subscripts)\cr}$$ ^^{display style}^^{text style}^^{script style}^^{scriptscript style} and four other ``^{cramped}'' styles that are almost the same except that exponents aren't raised quite so much. For brevity we shall refer to the eight styles as \begindisplay $\displaystyle D,\ D',\ T,\ T',\ S,\ S',\ \SS,\ \SS',$ \enddisplay where $D$ is display style, $D'$ is cramped display style, $T$~is text style, etc. \TeX\ also uses three different ^{sizes of type for mathematics}; they are called ^{text size}, ^{script size}, and ^{scriptscript size}. The normal way to typeset a formula with \TeX\ is to enclose it in dollar signs |$|$\,\ldots\,$|$|; this yields the formula in text style (style~$T$). Or you can enclose it in double dollar signs |$$|$\,\ldots\,$|$$|; this displays the formula in display style (style~$D$). The subformulas of a formula might, of course, be in different styles. Once you know the style, you can determine the size of type that \TeX\ will use: $$\everycr{\noalign{\penalty10000}} \halign{\indent#\hfil\qquad&#\hfil&\quad#\llap(like this)\hfil\cr If a letter is in style&then it will be set in\cr \noalign{\vskip 2pt} $D,D',T,T'$&text size&\cr $S,S'$&script size&\sevenrm\cr $\SS,\SS'$&scriptscript size&\fiverm\cr}$$ There is no ``$\it SSS$'' style or ``scriptscriptscript'' size; such tiny symbols would be even less readable than the scriptscript ones. Therefore \TeX\ stays with scriptscript size as the minimum: $$\halign{\indent\hbox to 1.3in{#\hfil}&\hbox to 1.2in{#\hfil}&#\hfil\cr In a formula&the superscript&and the subscript\cr of style&style is&style is\cr \noalign{\vskip 2pt} $D,T$&$S$&$S'$\cr $D',T'$&$S'$&$S'$\cr $S,\SS$&$\SS$&$\SS'$\cr $S',\SS'$&$\SS'$&$\SS'$\cr}$$ For example, if |x^{a_b}| is to be typeset in style $D$, then |a_b| will be set in style~$S$, and {\tt b}~in style~$\SS'$; the result is `$\displaystyle x^{a_b}$'. So far we haven't seen any difference between styles $D$ and $T$. Actually there is a slight difference in the positioning of exponents, although script size is used in each case: You get $\displaystyle x^2$~in $D$~style and $x^2$~in $T$~style and \vbox to 0pt{ \vss\hbox{$\displaystyle{\atop x^2}$}\kern0pt}~in $D'$ or $T'$~style---do you see the difference? But there is a big distinction between $D$ style and $T$ style when it comes to fractions: $$\halign{\indent\hbox to 1.3in{#\hfil}&\hbox to 1.2in{#\hfil}&#\hfil\cr In a formula&the style of the&and the style of the\cr $\alpha$|\over|$\,\beta$ of style&numerator $\alpha$ is&denominator $\beta$ is\cr \noalign{\vskip 2pt} $D$&$T$&$T'$\cr $D'$&$T'$&$T'$\cr $T$&$S$&$S'$\cr $T'$&$S'$&$S'$\cr $S,\SS$&$\SS$&$\SS'$\cr $S',\SS'$&$\SS'$&$\SS'$\cr}$$ ^^{numerator}^^{denominator} Thus if you type `|$1\over2$|' (in a text) you get $1\over2$, namely style $S$ over style~$S'$; but if you type `|$$1\over2$$|' you get $$1\over2$$ (a displayed formula), which is style $T$ over style $T'$. \danger While we're at it, we might as well finish the style rules: ^|\underline| does not change the style. ^{Math accents}, and the operations ^|\sqrt| and ^|\overline|, change uncramped styles to their cramped counterparts; for example, $D$ changes to $D'$, but $D'$ stays as it was. \dangerexercise State the style and size of each part of the formula $\displaystyle \sqrt{p_2^{e'}}$, assuming that the formula itself is in style~$D$. \answer Style $D'$ is used for the subformula $p_2^{e'}$, hence style~$S'$ is used for the superscript~$e'$ and the subscript~2, and style~$\SS'$ is used for the supersuperscript prime. The square root sign and the $p$ appear in text size; the 2 and the~$e$ appear in script size; and the $\prime$ is in scriptscript size. Suppose you don't like the style that \TeX\ selects by its automatic style rules. Then you can specify the style you want by typing ^|\displaystyle| or ^|\textstyle| or ^|\scriptstyle| or ^|\scriptscriptstyle|; the style that you select will apply until the end of the formula or subformula, or until you select another style. For example, `|$$n+\scriptstyle n+\scriptscriptstyle n.$$|' produces the display $$n+\scriptstyle n+\scriptscriptstyle n.$$ This is a rather silly example, but it does show that the plus signs get smaller too, as the style changes. \TeX\ puts no space around + signs in script styles. Here's a more useful example of style changes: Sometimes you need to typeset a ``^{continued fraction}'' made up of many other fractions, all of which are supposed to be in display style: $$a_0+{1\over\displaystyle a_1+ {\strut 1\over\displaystyle a_2+ {\strut 1\over\displaystyle a_3+ {\strut 1\over a_4}}}}$$ In order to get this effect, the idea is to type \begintt $$a_0+{1\over\displaystyle a_1+ {\strut 1\over\displaystyle a_2+ {\strut 1\over\displaystyle a_3+ {\strut 1\over a_4}}}}$$ \endtt (The control sequence ^|\strut| has been used to make the denominators taller; this is a refinement that will be discussed in Chapter~18. Our concern now is with the style commands.) \ Without the appearances of\/ |\strut| and |\displaystyle| in this formula, the result would be completely different: $$a_0+{1\over a_1+{1\over a_2+{1\over a_3+{1\over a_4}}}}$$ \danger These examples show that the numerator and denominator of a fraction are generally centered with respect to each other. If you prefer to have the numerator or denominator appear ^{flush left}, put `^|\hfill|' after it; or if you prefer ^{flush right}, put `|\hfill|' at the left. For example, if the first three appearances of `|1\over|' in the previous example are replaced by `|1\hfill\over|', you get the display $$a_0+{1\hfill\over\displaystyle a_1+ {\strut1\hfill\over\displaystyle a_2+ {\strut1\hfill\over\displaystyle a_3+ {\strut1\over a_4}}}}$$ (a format for continued fractions that many authors prefer). This works because |\hfill| stretches at a faster rate than the glue that is actually used internally by \TeX\ when it centers the numerators and denominators. \TeX\ has another operation `^|\atop|', which is like |\over| except that it leaves out the fraction line: \begindisplaymathdemo |$$x\atop y+2$$|&x\atop y+2\cr \endmathdemo The plain \TeX\ format in Appendix B also defines `^|\choose|', which is like |\atop| but it encloses the result in parentheses: \begindisplaymathdemo |$$n\choose k$$|&n\choose k\cr \endmathdemo It is called |\choose| because it's a common notation for the so-called ^{binomial coefficient} that tells how many ways there are to choose $k$~things out of $n$~things. You can't mix |\over| and |\atop| and |\choose| with each other. For example, `|$$n \choose k \over 2$$|' is illegal; you must use grouping, to get either `|$${n\choose k}\over2$$|' or `|$$n\choose{k\over2}$$|', i.e., \begindisplay $\displaystyle{{n\choose k}\over2}\qquad{\rm or}\qquad {n\choose{k\over2}}.$ \enddisplay The latter formula, incidentally, would look better as `|$$n\choose k/2$$|' or `|$$n\choose{1\over2}k$$|', yielding \begindisplay $\displaystyle{n\choose k/2}\qquad{\rm or}\qquad{n\choose{1\over2}k}.$ \enddisplay \medskip \exercise As alternatives to $\displaystyle{{n\choose k}\over2}$, discuss how you could obtain the two displays \begindisplay\abovedisplayskip=0pt\belowdisplayskip=0pt $\displaystyle {1\over2}{n\choose k} \qquad{\rm and}\qquad {\displaystyle{n\choose k}\over2}.$ \enddisplay \answer |$${1\over2}{n\choose k}$$|; |$$\displaystyle{n\choose k}\over2$$|. All of these braces are necessary. \bigbreak \exercise Explain how to specify the displayed formula $${p \choose 2}x^2 y^{p-2} - {1 \over 1-x}{1 \over 1-x^2}.$$ \answer |$${p \choose 2} x^2 y^{p-2} - {1 \over 1-x}{1 \over 1-x^2}.$$| \danger \TeX\ has a generalized version of\/ |\over| and |\atop| in which you specify the exact thickness of the line rule by typing `^|\above|\'. For example, \begintt $$\displaystyle{a\over b}\above1pt\displaystyle{c\over d}$$ \endtt will produce a ^{compound fraction} with a heavier ($1\pt$ thick) rule as its main bar: $${\displaystyle{a\over b}\above 1pt\displaystyle{c\over d}}.$$ This sort of thing occurs primarily in textbooks on elementary mathematics. \goodbreak Mathematicians often use the sign $\sum$ to stand for ``^{summation}'' and the sign $\int$ to stand for ``^{integration}.'' If you're a typist but not a mathematician, all you need to remember is that ^|\sum| stands for $\sum$ and ^|\int| for $\int$; these abbreviations appear in Appendix~F together with all the other symbols, in case you forget. Symbols like $\sum$ and $\int$ (and a few others like $\bigcup$ and $\prod$ and $\oint$ and~$\bigotimes$, all listed in Appendix~F) are called {\sl ^{large operators}}, ^^{collective signs, see large operators} ^^{sigma signs, see sum} and you type them just as you type ordinary symbols or letters. The difference is that \TeX\ will choose a {\sl larger\/} large operator in display style than it will in text style. For example, $$\halign{\indent#\hfil\qquad yields\qquad&$#\hfil$\qquad&#\hfil\cr |$\sum x_n$|&\sum x_n&($T$ style)\cr \noalign{\vskip3pt} |$$\sum x_n$$|&\displaystyle\sum x_n&($D$ style).\cr}$$ A displayed |\sum| usually occurs with ``^{limits},'' i.e., with subformulas that are to appear above and below it. You type limits just as if they were superscripts and subscripts; for example, if you want $$\sum_{n=1}^m$$ you type either `|$$\sum_{n=1}^m$$|' or `|$$\sum^m_{n=1}$$|'. According to the normal conventions of mathematical typesetting, \TeX\ will change this to `$\sum_{n=1}^m$' (i.e., without limits) if it occurs in text style rather than in display style. Integrations are slightly different from summations, in that the superscripts and subscripts are not set as limits even in display style: $$\halign{\indent\hbox to2.3in{#\hfil}\hbox to.6in{yields\hfil}& $#\hfil$\qquad&#\hfil\cr |$\int_{-\infty}^{+\infty}$|&\int_{-\infty}^{+\infty}&($T$ style)\cr \noalign{\vskip3pt} |$$\int_{-\infty}^{+\infty}$$|&\displaystyle\int_{-\infty}^{+\infty}& ($D$ style).\cr}$$ \danger Some printers prefer to set limits above and below $\int$ signs; this takes more space on the page, but it gives a better appearance if the subformulas are complex, because it keeps them out of the way of the rest of the formula. Similarly, limits are occasionally desirable in text style or script style; but some printers prefer not to set limits on displayed $\sum$ signs. You can change \TeX's convention by simply typing `^|\limits|' or `^|\nolimits|' immediately after the large operator. For example, $$\halign{\indent\hbox to2.3in{#\hfil}\hbox to.6in{yields\hfil}& $\displaystyle{#}$\hfil\cr |$$\int\limits_0^{\pi\over2}$$|&\int\limits_0^{\pi\over2}\cr \noalign{\vskip 4pt} |$$\sum\nolimits_{n=1}^m$$|&\sum\nolimits_{n=1}^m\cr}$$ \ddanger If you say `|\nolimits\limits|' (presumably because some macro like |\int| specifies |\nolimits|, but you do want them), the last word takes precedence. There's also a command `^|\displaylimits|' that can be used to restore \TeX's normal conventions; i.e., the limits will be displayed only in styles $D$ and $D'$. \danger Sometimes you need to put two or more rows of limits under a large operator; you can do this with `^|\atop|'. For example, if you want the displayed formula $$\sum_{\scriptstyle0\le i\le m\atop\scriptstyle0\^|\right|\ \enddisplay and \TeX\ will typeset the subformula, putting the specified delimiters at the left and the right. The size of the delimiters will be just big enough to cover the subformula. For example, in the display \beginlongdisplaymathdemo |$$1+\left(1\over1-x^2\right)^3$$|&1+\left(1\over1-x^2\right)^3\cr \endmathdemo \TeX\ has chosen |\biggl(| and |\biggr)|, because smaller delimiters would be too small for this particular fraction. A simple formula like `|$\left(x\right)$|' yields just `$\left(x\right)$'; thus, |\left| and |\right| sometimes choose delimiters that are smaller than |\bigl| and |\bigr|. Whenever you use |\left| and |\right| they must pair up with each other, just as braces do in groups. You can't have |\left| in one formula and |\right| in another, nor are you allowed to type things like `|\left(...{...\right)...}|' or `|\left(...\begingroup...\right)...\endgroup|'. This restriction makes sense, because \TeX\ needs to typeset the subformula that appears between |\left| and |\right| before it can decide how big to make the delimiters. But it is worth explicit mention here, because you do {\sl not\/} have to match ^{parentheses} and ^{brackets}, etc., ^^{crotchets, see brackets} when you are not using |\left| and |\right|: \TeX\ will not complain if you input a formula like `|$[0,1)$|' or even `|$)($|' or just `|$)$|'.\ (And it's a good thing \TeX\ doesn't, for such unbalanced formulas occur surprisingly often in mathematics papers.) \ Even when you do use |\left| and |\right|, \TeX\ doesn't look closely at the particular delimiters that you happen to choose; thus, you can type strange things like `|\left)|' and/or `|\right(|' if you know what you're doing. Or even if you don't. The |\over| operation in the example displayed above does not involve the `|1+|' at the beginning of the formula; this happens because |\left| and |\right| have the function of ^{grouping}, in addition to their function of delimiter-making. Any definitions that you happen to make between |\left| and |\right| will be local, as if braces had appeared around the enclosed subformula. \exercise Use |\left| and |\right| to typeset the following display (with ^|\phi| for $\phi$): $$\pi(n)=\sum_{k=2}^n\left\lfloor\phi(k)\over k-1\right\rfloor.$$ \answer |$$\pi(n)=\sum_{k=2}^n\left\lfloor\phi(k)\over k-1\right\rfloor.$$| At this point you are probably wondering why you should bother learning about |\bigl| and |\bigr| and their relatives, when |\left| and |\right| are there to calculate sizes for you automatically. Well, it's true that |\left| and |\right| are quite handy, but there are at least three situations in which you will want to use your own wisdom when selecting the proper delimiter size: \ (1)~Sometimes |\left| and |\right| choose a smaller delimiter than you want. For example, we used |\bigl| and |\bigr| to produce $\bigl\vert\vert x\vert- \vert y\vert\bigr\vert$ in one of the previous illustrations; |\left| and |\right| don't make things any bigger than necessary, so `|$\left|\||\left|\||x\right|\||-\left|\||y\right|\||\right|\||$|' yields only `$\left\vert \left\vert x\right\vert -\left\vert y\right\vert \right\vert$'. \ (2)~Sometimes |\left| and |\right| choose a larger delimiter than you want. This happens most frequently when they enclose a large operator in a display; for example, compare the following two formulas: \beginlongdisplaymathdemo \noalign{\vskip 6pt} |$$\left( \sum_{k=1}^n A_k \right)$$|&\left( \sum_{k=1}^n A_k \right)\cr \noalign{\vskip 3pt} |$$\biggl( \sum_{k=1}^n A_k \biggr)$$|&\biggl( \sum_{k=1}^n A_k \biggr)\cr \endmathdemo The rules of\/ |\left| and |\right| cause them to enclose the ^|\sum| together with its ^{limits}, but in special cases like this it looks better to let the limits hang out a~bit; |\bigg| delimiters are better here. \ (3)~Sometimes you need to break a huge displayed formula into two or more separate lines, and you want to make sure that its opening and closing delimiters have the same size; but you can't use |\left| on the first line and |\right| on the last, since |\left| and |\right| must occur in pairs. The solution is to use |\Biggl| (say) on the first line and |\Biggr| on the last. \danger Of course, one of the advantages of\/ |\left| and |\right| is that they can make arbitrarily large delimiters---much bigger than |\biggggg|! The slashes and angle brackets do have a maximum size, however; if you ask for really big versions of those symbols you will get the largest ones available. \exercise Prove that you have mastered delimiters: Coerce \TeX\ into producing the formula $$\pi(n)=\sum_{m=2}^n\left\lfloor\biggl(\sum_{k=1}^{m-1}\bigl\lfloor (m/k)\big/\lceil m/k\rceil\bigr\rfloor\biggr)^{-1}\right\rfloor.$$ \answer |$$\pi(n)=\sum_{m=2}^n\left\lfloor\biggl(\sum_{k=1}^{m-1}\bigl| \hfil\break |\lfloor(m/k)\big/\lceil m/k\rceil\bigr\rfloor\biggr)^{-1}\right\rfloor.$$| \danger If you type `|.|'\ after |\left| or |\right|, instead of specifying one of the basic delimiters, you get a so-called ^{null delimiter} (which is blank). Why on earth would anybody want that, you may ask. Well, you sometimes need to produce formulas that contain only one large delimiter. For example, the display $$\vert x\vert=\cases{x,&if $x\ge0$\cr -x,&if $x<0$\cr}$$ has a `$\{$' but no `$\}$'. It can be produced by a construction of the form \begindisplay |$$|\||x|\||=\left\{ ... \right.$$| \enddisplay Chapter 18 explains how to fill in the `\hbox{|...|}' to finish this construction; let's just notice for now that the `|\right.|'\ makes it possible to have an invisible right delimiter to go with the visible left brace. \ddanger A null delimiter isn't completely void; it is an empty box whose width is a \TeX\ parameter called ^|\nulldelimiterspace|. We will see later that null delimiters are inserted next to fractions. Plain \TeX\ sets |\nulldelimiterspace=1.2pt|. You can type `|<|' or `|>|' as convenient abbreviations for ^|\langle| and ^|\rangle|, when \TeX\ is looking for a delimiter. For example, `|\bigl<|' is equivalent to `|\bigl\langle|', and `|\right>|' is equivalent to `|\right\rangle|'. Of course `|<|' and `|>|' ordinarily produce the ^{less-than} and ^{greater-than} relations `${<}\,{>}$', which are quite different from ^{angle brackets} `$\langle\,\rangle$'. \danger Plain \TeX\ also makes available a few more delimiters, which were not listed in the basic set of~22 because they are sort of special. The control sequences ^|\arrowvert|, ^|\Arrowvert|, and ^|\bracevert| produce delimiters made from the repeatable parts of the vertical arrows, double vertical arrows, and large braces, respectively, without the arrowheads or the curly parts of the braces. They produce results similar to ^|\vert| or ^|\Vert|, but they are surrounded by more white space and they have a different weight. You can also use ^|\lgroup| and ^|\rgroup|, which are constructed from braces without the middle parts; and ^|\lmoustache| and ^|\rmoustache|, ^^{moustaches} which give you the top and bottom halves of large braces. For example, here are the |\Big| and |\bigg| versions of\/ |\vert|, |\Vert|, and these seven special delimiters: $$\halign{\indent$#\hfil$\cr \ldots\Big\vert\ldots\Big\Vert \ldots\Big\arrowvert\ldots\Big\Arrowvert\ldots\Big\bracevert \ldots\Big\lgroup\ldots\Big\rgroup\ldots\Big\lmoustache\ldots\Big\rmoustache \ldots\,;\cr \noalign{\smallskip} \ldots\bigg\vert\ldots\bigg\Vert \ldots\bigg\arrowvert\ldots\bigg\Arrowvert\ldots\bigg\bracevert \ldots\bigg\lgroup\ldots\bigg\rgroup\ldots\bigg\lmoustache\ldots\bigg\rmoustache \ldots\,.\cr}$$ Notice that |\lgroup| and |\rgroup| are rather like bold parentheses, with sharper bends at the corners; this makes them attractive for certain large displays. But you cannot use them exactly like parentheses, because they are available only in large sizes (|\Big|~or~more). \ddanger Question: What happens if a ^{subscript} or ^{superscript} follows a large delimiter? Answer:~That's a good question. After a |\left| delimiter, it is the first subscript or superscript of the enclosed subformula, so it is effectively preceded by |{}|. After a |\right| delimiter, it is a subscript or superscript of the entire |\left...\right| subformula. And after a |\bigl| or |\bigr| or |\bigm| or |\big| delimiter, it applies only to that particular delimiter. Thus, `|\bigl(_2|' works quite differently from `|\left(_2|'. \danger If you look closely at the examples of math typesetting in this chapter, you will notice that large parentheses and brackets are symmetric with respect to an invisible horizontal line that runs a little bit above the ^{baseline}; when a delimiter gets larger, its height and depth both grow by the same amount. This horizontal line is called the {\sl^{axis}\/} of the formula; for example, a formula in the text of the present paragraph would have an axis at this level: $\hskip 2em\over$. The bar line in every fraction is centered on the axis, regardless of the size of the numerator or denominator. \danger Sometimes it is necessary to create a special box that should be centered vertically with respect to the axis. \ (For example, the `$\vert x\vert=\bigl\{\,\ldots$' example above was done with such a box.) \ \TeX\ provides a simple way to do this: You just say \begindisplay |\vcenter{|\|}| \enddisplay and the vertical mode material will be packed into a box just as if ^|\vcenter| had been ^|\vbox|. Then the box will be raised or lowered until its top edge is as far above the axis as the bottom edge is below. \ddanger The concept of ``axis'' is meaningful for \TeX\ only in math formulas, not in ordinary text; therefore \TeX\ allows you to use |\vcenter| only in math mode. If you really need to center something vertically in horizontal mode, the solution is to say `|$\vcenter{...}$|'. \ (Incidentally, the constructions `|\vcenter| |to|\' and `|\vcenter| |spread|\' are legal too, in math mode; vertical glue is always set by the rules for |\vbox| in Chapter~12. But |\vcenter| by itself is usually sufficient.) \danger Any box can be put into a formula by simply saying ^|\hbox| or |\vbox| or ^|\vtop| or ^|\box| or ^|\copy| in the normal way, even when you are in math mode. Furthermore you can use ^|\raise| or ^|\lower|, as if you were in horizontal mode, and you can insert vertical rules with ^|\vrule|. Such constructions, like |\vcenter|, produce boxes that can be used like ordinary symbols in math formulas. \ddanger Sometimes you need to make up your own symbols, when you run across something unusual that doesn't occur in the fonts. If the new symbol occurs only in one place, you can use |\hbox| or |\vcenter| or something to insert exactly what you want; but if you are defining a macro for general use, you may want to use different constructions in different styles. \TeX\ has a special feature called ^|\mathchoice| that comes to the rescue in such situations: You write \begindisplay |\mathchoice{|\|}{|\|}{|\|}{|\|}| \enddisplay where each \ specifies a subformula. \TeX\ will choose the first subformula in style $D$ or~$D'$, the second in style $T$ or~$T'$, the third in style $S$ or~$S'$, the fourth in style $\SS$ or $\SS'$. \ (\TeX\ actually typesets all four subformulas, before it chooses the final one, because the actual style is not always known at the time a |\mathchoice| is encountered; for example, when you type `|\over|' you often change the style of everything that has occurred earlier in the formula. Therefore |\mathchoice| is somewhat expensive in terms of time and space, and you should use it only when you're willing to pay the price.) \ddangerexercise Guess what output is produced by the following commands: \begintt \def\puzzle{{\mathchoice{D}{T}{S}{SS}}} $$\puzzle{\puzzle\over\puzzle^{\puzzle^\puzzle}}$$ \endtt \answer A displayed formula equivalent to |$${D}{{T}\over{T}^{{S}^{SS}}}$$|. \ddangerexercise Devise a `^|\square|' macro that produces a \def\sqr#1#2{{\vcenter{\vbox{\hrule height.#2pt \hbox{\vrule width.#2pt height#1pt \kern#1pt \vrule} \hrule height.#2pt}}}}% `$\,\sqr34\,$' for use in math formulas. The box should be symmetrical with respect to the axis, and its inside dimensions should be $3\pt$ in display and text styles, $2.1\pt$ in script styles, and $1.5\pt$ in scriptscript styles. The rules should be $0.4\pt$ thick in display and text styles, $0.3\pt$ thick otherwise. \answer |\def\sqr#1#2{{\vcenter{\vbox{\hrule height.#2pt|\parbreak | \hbox{\vrule width.#2pt height#1pt \kern#1pt|\parbreak | \vrule width.#2pt}|\parbreak | \hrule height.#2pt}}}}|\parbreak |\def\square{\mathchoice\sqr34\sqr34\sqr{2.1}3\sqr{1.5}3}| \ddanger Plain \TeX\ has a macro called ^|\mathpalette| that is useful for |\mathchoice| constructions; `|\mathpalette\a{xyz}|' expands to the four-pronged array of choices `|\mathchoice|\stretch|{\a|\stretch|\displaystyle|\stretch|{xyz}}|\stretch |...|\stretch|{\a|\stretch|\scriptscriptstyle|\stretch|{xyz}}|\stretch'. Thus the first argument to |\mathpalette| is a control sequence whose first argument is a style selection. Appendix~B contains several examples that show how |\mathpalette| can be applied. \ (See in particular the definitions of\/ |\phantom|, |\root|, and |\smash|; the ^{congruence sign} ^|\cong| ($\cong$) is also constructed from $=$ and $\sim$ using |\mathpalette|.) ^^{constructing new math symbols} ^^{math symbols, construction of} \ddanger At the beginning of this chapter we discussed the commands |\over|, |\atop|, |\choose|, and |\above|. These are special cases of \TeX's ``^{generalized fraction}'' feature, which includes also the three primitives \begindisplay |\overwithdelims|\\\cr |\atopwithdelims|\\\cr |\abovewithdelims|\\\\cr \enddisplay The third of these is the most general, as it encompasses all of the other generalized fractions: ^|\overwithdelims| uses a ^{fraction} bar whose thickness is the default for the current size, and ^|\atopwithdelims| uses an invisible fraction bar whose thickness is zero, while ^|\abovewithdelims| uses a bar whose thickness is specified explicitly. \TeX\ places the immediately preceding subformula (the ^{numerator}) over the immediately following subformula (the ^{denominator}), separated by a bar line of the desired thickness; then it puts \ at the left and \ at the right. For example, `^|\choose|' is equivalent to `|\atopwithdelims()|'. If you define |\legendre| to be `|\overwithdelims()|', you can typeset the ^{Legendre symbol} \def\legendre{\overwithdelims()}% `$a\legendre b$' by saying `|{a\legendre b}|'. The size of the surrounding delimiters depends only on the style, not on the size of the fractions; larger delimiters are used in styles $D$ and~$D'$ (see Appendix~G\null). The simple commands ^|\over|, ^|\atop|, and ^|\above| are equivalent to the corresponding `|withdelims|' commands when the delimiters are null; for example, `|\over|' is an abbreviation for `|\overwithdelims..|'. \def\euler{\atopwithdelims<>} \ddangerexercise Define a control sequence |\euler| so that the ^{Eulerian number} $n\euler k$ will be produced when you type `|{n\euler k}|' in a formula. \answer|\def\euler{\atopwithdelims<>}|. \ddanger Appendix G explains exactly how \TeX\ computes the desired size of delimiters for |\left| and~|\right|. The general idea is that delimiters are vertically centered with respect to the ^{axis}; hence, if we want to cover a subformula between |\left| and |\right| that extends $y_1$~units above the axis and $y_2$~units below, we need to make a delimiter whose height plus depth is at least $y$~units, where $y=2\max(y_1,y_2)$. It is usually best not to cover the formula completely, however, but just to come close; so \TeX\ allows you to specify two parameters, the ^|\delimiterfactor|~$f$ (an~integer) and the ^|\delimitershortfall|~$\delta$ (a~dimension). The minimum delimiter size is taken to be at least $y\cdot f/1000$, and at least $y-\delta$. Appendix~B sets $f=901$ and $\delta=5\pt$. Thus, if $y=30\pt$, the plain \TeX\ format causes the delimiter to be more than $27\pt$ tall; if $y=100\pt$, the corresponding delimiter will be at least $95\pt$ tall. \danger So far we have been discussing the rules for typing math formulas, but we haven't said much about how \TeX\ actually goes about converting its input into lists of boxes and glue. Almost all of the control sequences that have been mentioned in Chapters 16 and~17 are ``high level'' features of the plain \TeX\ format; they are not built into \TeX\ itself. Appendix~B defines those control sequences in terms of more primitive commands that \TeX\ actually deals with. For example, `|\choose|' is an abbreviation for `|\atopwithdelims()|'; Appendix~B not only introduces |\choose|, it also tells \TeX\ where to find the delimiters |(| and~|)| in various sizes. The plain \TeX\ format defines all of the special characters like |\alpha| and~|\mapsto|, all of the special accents like |\tilde| and~|\widehat|, all of the large operators like |\sum| and~|\int|, and all of the delimiters like |\lfloor| and~|\vert|. Any of these things can be redefined, in order to adapt \TeX\ to other mathematical styles and/or to other fonts. \danger The remainder of this chapter discusses the low-level commands that \TeX\ actually obeys behind the scenes. Every paragraph on the next few pages is marked with double dangerous bends, so you should skip to Chapter~18 unless you are a glutton for \TeX nicalities. \ninepoint \ddanger All characters that are typeset in math mode belong to one of sixteen {\sl^{families} of fonts}, numbered internally from 0 to~15. Each of these families consists of three fonts: one for text size, one for script size, and one for scriptscriptsize. The commands ^|\textfont|, ^|\scriptfont|, and ^|\scriptscriptfont| are used to specify the members of each family. For example, ^{family~0} in the plain \TeX\ format is used for roman letters, and Appendix~B contains the instructions \begintt \textfont0=\tenrm \scriptfont0=\sevenrm \scriptscriptfont0=\fiverm \endtt to set up this family: The 10-point roman font (^|\tenrm|) is used for normal symbols, 7-point roman (^|\sevenrm|) is used for subscripts, and 5-point roman (^|\fiverm|) is used for sub-subscripts. Since there are up to 256~characters per font, and 3~fonts per family, and 16~families, \TeX\ can access up to 12,288 characters in any one formula (4096 in~each of the three sizes). Imagine that. \ddanger A definition like |\textfont|\|=|\ is local to the group that contains it, so you can easily change family membership from one set of conventions to another and back again. Furthermore you can put any font into any family; for example, the command \begintt \scriptscriptfont0=\scriptfont0 \endtt makes sub-subscripts in family~0 the same size as the subscripts currently are. \TeX\ doesn't check to see if the families are sensibly organized; it just follows instructions. \ (However, fonts cannot be used in families 2 and~3 unless they contain a certain number of special parameters, as we shall see later.) \ Incidentally, \TeX\ uses ^|\nullfont|, which contains no characters, for each family member that has not been defined. \ddanger During the time that a math formula is being read, \TeX\ remembers each symbol as being ``character position so-and-so in family number such-and-such,'' but it does not take note of what fonts are actually in the families until reaching the end of the formula. Thus, if you have loaded a font called |\Helvetica| that contains Swiss-style numerals, and if you say something like \begintt $\textfont0=\tenrm 9 \textfont0=\Helvetica 9$ \endtt you will get two 9's in font |\Helvetica|, assuming that \TeX\ has been set up to take 9's from family~0. The reason is that |\textfont0| is~|\Helvetica| at the end of the formula, and that's when it counts. On the other hand, if you say \begintt $\textfont0=\tenrm 9 \hbox{$9\textfont0=\Helvetica$}$ \endtt the first 9 will be from |\tenrm| and the second from |\Helvetica|, because the formula in the hbox will be typeset before it is incorporated into the surrounding formula. \ddangerexercise If you say `|${\textfont0=\Helvetica 9}$|', what font will be used for the~9? \answer The |\textfont0| that was current at the beginning of the formula will be used, because this redefinition is local to the braces. \ (It would be a different story if `^|\global||\textfont|' had appeared instead; that would have changed the meaning of\/ |\textfont0| at all levels.) \ddanger Every ^{math character} is given an identifying code number between 0 and~4095, obtained by adding 256~times the family number to the position number. This is easily expressed in ^{hexadecimal notation}, using one hexadecimal digit for the family and two for the character; for example, \hex{24A} stands for character~\hex{4A} in family~2. Each character is also assigned to one of eight classes, ^^{classes of math characters, table} ^^{math codes} ^^{table of ...} numbered 0 to~7, as follows: $$\halign{\indent#\hfil&\quad#\hfil&\quad#\hfil& \hskip4em#\hfil&\quad#\hfil&\quad#\hfil\cr \it \kern-2pt Class&\it Meaning&\kern-2pt\it Example& \it \kern-2pt Class&\it Meaning&\kern-2pt\it Example\cr \noalign{\vskip2pt} 0&Ordinary&|/|& 4&Opening&|(|\cr 1&Large operator&|\sum|& 5&Closing&|)|\cr 2&Binary operation&|+|& 6&Punctuation&|,|\cr 3&Relation&|=|& 7&Variable family&|x|\cr }$$ ^^{large operator}^^{binary operation}^^{relation}^^{opening}^^{closing} ^^{punctuation}^^{variable family} Classes 0 to 6 tell what ``part of speech'' the character belongs to, in math-printing language; class~7 is a special case discussed below. The class number is multiplied by 4096 and added to the character number, and this is the same as making it the leading digit of a four-digit hexadecimal number. For example, Appendix~B defines |\sum| to be the math character \hex{1350}, meaning that it is a large operator (class~1) found in position \hex{50} of family~3. \ddangerexercise The ^|\oplus| and ^|\bullet| symbols ($\oplus$ and $\bullet$) are binary operations that appear in positions 8 and~15 (decimal) of family~2, when the fonts of plain~\TeX\ are being used. Guess what their math character codes are. \ (This is too easy.) \answer \hex{2208} and \hex{220F}. \ddanger Class 7 is a special case that allows math symbols to change families. It behaves exactly like class~0, except that the specified family is replaced by the current value of an integer parameter called ^|\fam|, provided that |\fam| is a legal family number (i.e., if it lies between 0 and~15). \TeX\ automatically sets |\fam=-1| whenever math mode is entered; therefore class~7 and class~0 are equivalent unless |\fam| has been given a new value. Plain \TeX\ changes |\fam| to~0 when the user types `^|\rm|'; this makes it convenient to get roman letters in formulas, as we will see in Chapter~18, since letters belong to class~7. \ (The control sequence |\rm| is an abbreviation for `|\fam=0 \tenrm|'; thus, |\rm| causes |\fam| to become zero, and it makes |\tenrm| the ``^{current font}.'' In horizontal mode, the |\fam| value is irrelevant and the current font governs the typesetting of letters; but in math mode, the current font is irrelevant and the |\fam| value governs the letters. The current font affects math mode only if\/ |\|\] is used ^^{control space} or if dimensions are given in ^|ex| or ^|em| units; it also has an effect if an |\hbox| appears inside a formula, since the contents of an hbox are typeset in horizontal mode.) \ddanger The interpretation of characters in math mode is defined by a table of~256 ``mathcode'' values; these table entries can be changed by the ^|\mathcode| command, just as the category codes are changed by ^|\catcode| (see Chapter~7). Each mathcode specifies class, family, and character position, as described above. For example, Appendix~B contains the commands \begintt \mathcode`<="313C \mathcode`*="2203 \endtt which cause \TeX\ to treat the character `|<|' in math mode as a relation ^^{less than} (class~3) found in position \hex{3C} of family~1, and to treat an ^{asterisk} `|*|' as a binary operation found in position~3 of family~2. The initial value of\/ |\mathcode`b| is \hex{7162}; thus, |b|~is character \hex{62} in ^{family~1} (italics), and its family will vary with |\fam|. \ (|INITEX| starts out with |\mathcode|$\,x=x$ for all characters~$x$ that are neither letters nor digits. The ten digits have |\mathcode|$\,x=x+\hbox{\hex{7000}}$; the 52 letters have |\mathcode|$\,x=x+\hbox{\hex{7100}}$.) \ \TeX\ looks at the mathcode only when it is typesetting a character whose catcode is 11~(letter) or 12~(other), or when it encounters a character that is given explicitly as ^|\char|\. \ddanger A |\mathcode| can also have the special value \hex{8000}, which causes the character to behave as if it has catcode~13 (active). Appendix~B uses this feature to make |'| ^^{apostrophe} expand to |^{|^|\prime||}| in a slightly tricky way. The mathcode of |'| does not ^^{active math character} interfere with the use of |'| in ^{octal} constants. \ddanger The mathcode table allows you to refer indirectly to any character in any family, with the touch of a single key. You can also specify a math character code directly, by typing ^|\mathchar|, which is analogous to ^|\char|. For example, the command `|\mathchar"1ABC|' specifies a character of class~1, family~10 (\hex A), and position \hex{BC}. A~hundred or so definitions like \begintt \def\sum{\mathchar"1350 } \endtt would therefore suffice to define the special symbols of plain \TeX\null. But there is a better way: \TeX\ has a primitive command ^|\mathchardef|, which relates to |\mathchar| just as ^|\chardef| does to |\char|. Appendix~B has a hundred or so definitions like \begintt \mathchardef\sum="1350 \endtt to define the special symbols. A |\mathchar| must be between 0 and 32767 (\hex{7FFF}). \ddanger A character of class~1, i.e., a ^{large operator} like |\sum|, will be vertically centered with respect to the axis when it is typeset. Thus, the large operators can be used with different sizes of type. This vertical adjustment is not made for symbols of the other classes. \ddanger \TeX\ associates classes with subformulas as well as with individual characters. Thus, for example, you can treat a complex construction as if it were a binary operation or a relation, etc., if you want to. The commands ^|\mathord|, ^|\mathop|, ^|\mathbin|, ^|\mathrel|, ^|\mathopen|, ^|\mathclose|, and ^|\mathpunct| are used for this purpose; each of them is followed either by a single character or by a subformula in braces. For example, |\mathopen\mathchar"1234| is equivalent to |\mathchar"4234|, because |\mathopen| forces class~4 (opening). In the formula `|$G\mathbin:H$|', the ^{colon} is~treated as a binary operation. And Appendix~B constructs large opening symbols by defining ^|\bigl||#1| to be an abbreviation for \begintt \mathopen{\hbox{$\left#1 ...\right.$}} \endtt There's also an eighth classification, ^|\mathinner|, which is not normally used for individual symbols; fractions and ^|\left||...|^|\right| constructions are treated as ``inner'' subformulas, which means that they will be surrounded by additional space in certain circumstances. All other subformulas are generally treated as ordinary symbols, whether they are formed by |\overline| or |\hbox| or |\vcenter| or by simply being enclosed in braces. Thus, |\mathord| isn't really a necessary part of the \TeX\ language; instead of typing `|$1\mathord,234$|' you can get the same effect from `|$1{,}234$|'. \ddangerexercise Commands like |\mathchardef\alpha="010B| are used in Appendix~B to define the lowercase ^{Greek} letters. Suppose that you want to extend plain \TeX\ by putting ^{boldface math italic} letters in family~9, analogous to the normal math italic letters in family~1. \ (Such fonts aren't available in stripped down versions of \TeX, but let's assume that they exist.) \ Assume that the control sequence |\bmit| has been defined as an abbreviation for `|\fam=9|'; hence `|{\bmit b}|' will give a boldface math italic~|b|. What change to the definition of\/ |\alpha| will make |{\bmit\alpha}| produce a boldface~alpha? \checkequals\bmiexno\exno \answer |\mathchardef\alpha="710B|. Incidentally, |{\rm\alpha}| will then give a spurious result, because character position \hex{0B} of roman fonts does not contain an alpha; you should warn your users about what characters they are allowed to type under the influence of special conventions like ^|\rm|. \ddanger ^{Delimiters} are specified in a similar but more complicated way. Each character has not only a~|\catcode| and a~|\mathcode| but also a~^|\delcode|, which is either negative (for characters that should not act as delimiters) or less than \hex{1000000}. In other words, nonnegative delcodes consist of six hexadecimal digits. The first three digits specify a ``small'' variant of the delimiter, and the last three specify a ``large'' variant. For example, the command \begintt \delcode`x="123456 \endtt means that if the letter |x| is used as a delimiter, its small variant is found in position \hex{23} of family~1, and its large variant is found in position \hex{56} of family~4. If the small or large variant is given as |000|, however (position~0 of ^{family~0}), that variant is ignored. \TeX\ looks at the delcode when a character follows ^|\left| or ^|\right|, or when a character follows one of the ^|withdelims| commands; a negative delcode leads to an error message, but otherwise \TeX\ finds a suitable delimiter by first trying the small variant and then the large. \ (Appendix~G discusses this process in more detail.) \ For example, Appendix~B contains the commands \begintt \delcode`(="028300 \delcode`.=0 \endtt which specify that the small variant of a left parenthesis is found in position \hex{28} of family~0, and that the large variant is in position~0 of family~3; also, a period has no variants, hence `|\left.|'\ will produce a ^{null delimiter}. There actually are several different left parenthesis symbols in family~3; the smallest is in position~0, and the others are linked together by information that comes with the font. All delcodes are~$-1$ until they are changed by a |\delcode| command. \ddangerexercise Appendix~B defines |\delcode`<| so that there is a shorthand notation for ^{angle brackets}. Why do you think Appendix~B doesn't go further and define |\delcode`{|? \answer If\/ |\delcode`{| were set to some nonnegative delimiter code, you would get no error message when you wrote something like `|\left{|'. This would be bad because strange effects would happen when certain subformulas were given as arguments to macros, or when they appeared in alignments. But it has an even worse defect, because a user who gets away with `|\left{|' is likely to try also `|\bigl{|', which fails miserably. \ddanger A delimiter can also be given directly, as `^|\delimiter|\'. In this case the number can be as high as \hex{7FFFFFF}, i.e., seven hexadecimal digits; the leading digit specifies a class, from 0 to~7, as in a |\mathchar|. For example, Appendix~B contains the definition \begintt \def\langle{\delimiter"426830A } \endtt and this means that ^|\langle| is an opening (class 4) whose small variant is \hex{268} and whose large variant is \hex{30A}. When |\delimiter| appears after |\left| or |\right|, the class digit is ignored; but when |\delimiter| occurs in other contexts, i.e., when \TeX\ isn't looking for a delimiter, the three rightmost digits are dropped and the remaining four digits act as a |\mathchar|. For example, the expression `|$\langle x$|' is treated as if it were `|$\mathchar"4268 x$|'. \ddangerexercise What goes wrong if you type `|\bigl\delimiter"426830A|'\thinspace? \answer Since |\bigl| is defined as a macro with one parameter, it gets just `|\delimiter|' as the argument. You have to write `|\bigl{\delimiter"426830A}|' to make this work. On the other hand, |\left| will balk if the following character is a left brace. Therefore it's best to have control sequence names for all delimiters. \ddanger Granted that these numeric conventions for |\mathchar| and |\delimiter| are not beautiful, they sure do pack a lot of information into a small space. That's why \TeX\ uses them for low-level definitions inside formats. Two other low-level primitives also deserve to be mentioned: ^|\radical| and ^|\mathaccent|. Plain \TeX\ makes ^{square root signs} and math accents available by giving the commands \begintt \def\sqrt{\radical"270370 } \def\widehat{\mathaccent"362 } \endtt and several more like them. The idea is that |\radical| is followed by a delimiter code and |\mathaccent| is followed by a math character code, so that \TeX\ knows the family and character positions for the symbols used in radical and accent constructions. Appendix~G gives precise information about the positioning of these characters. By changing the definitions, \TeX\ could easily be extended so that it would typeset a variety of different radical signs and a variety of different accent signs, if such symbols were available in the fonts. ^^{surd signs, see radical} \ddanger Plain \TeX\ uses ^{family~1} for math italic letters, ^{family~2} for ordinary math symbols, and ^{family~3} for large symbols. \TeX\ insists that ^^{math fonts} the fonts in families 2 and~3 have special ^|\fontdimen| parameters, which govern mathematical spacing according to the rules in Appendix~G\null; the ^|cmsy| and ^|cmex| ^{symbol fonts} have these parameters, so their assignment to families 2 and~3 is almost mandatory. \ (There is, however, a way to modify the parameters of any font, using the ^|\fontdimen| command.) \ ^|INITEX| initializes the mathcodes of all ^{letters} |A| to~|Z| and |a| to~|z| so that they are symbols of class~7 and family~1; that's why it is natural to use family~1 for math italics. Similarly, the digits |0| to~|9| are class~7 and family~0. None of the other families is treated in any special way by \TeX. Thus, for example, plain \TeX\ puts ^{text italic} in family~4, slanted roman in family~5, bold roman in family~6, and typewriter type in family~7, but any of these numbers could be switched around. There is a macro ^|\newfam|, analogous to |\newbox|, that will assign symbolic names to families that aren't already used. \ddanger When \TeX\ is in horizontal mode, it is making a horizontal list; in vertical mode, it is making a vertical list. Therefore it should come as no great surprise that \TeX\ is making a ^{math list} when it is in ^{math mode}. The contents of horizontal lists were explained in Chapter~14, and the contents of vertical lists were explained in Chapter~15; it's time now to describe what math lists are made of. Each item in a math list is one of the following types of things:\enddanger \smallskip \item\bull an ^{atom} (to be explained momentarily); \item\bull horizontal material (a rule or discretionary or penalty or ``whatsit''); \item\bull vertical material (from |\mark| or |\insert| or |\vadjust|); \item\bull a glob of ^{glue} (from |\hskip| or |\mskip| or |\nonscript|); \item\bull a ^{kern} (from |\kern| or |\mkern|); \item\bull a ^{style change} (from |\displaystyle|, |\textstyle|, etc.); \item\bull a ^{generalized fraction} (from |\above|, |\over|, etc.); \item\bull a ^{boundary} (from |\left| or |\right|); \item\bull a four-way ^{choice} (from ^|\mathchoice|). \ddanger The most important items are called {\sl atoms}, and they have three parts: a {\sl^{nucleus}}, a {\sl^{superscript}}, and a {\sl^{subscript}}. For example, if you type \begintt (x_i+y)^{\overline{n+1}} \endtt in math mode, you get a math list consisting of five atoms: $($, $x_i$, $+$, $y$, and~$)^{\overline{n+1}}$. The nuclei of these atoms are $($, $x$, $+$, $y$, and~$)$; the subscripts are empty except for the second atom, which has subscript~$i$; the superscripts are empty except for the last atom, whose superscript is~$\overline{n+1}$. This superscript is itself a math list consisting of one atom, whose nucleus is~$n+1$; and that nucleus is a math list consisting of three atoms. \ddanger There are thirteen kinds of atoms, each of which might act differently in a formula; for example, `$($' is an Open atom because ^^{atomic types, table} it comes from an opening. Here is a complete list of the different kinds: $$\halign{\indent#\hfil&\enskip#\hfil\cr Ord&is an ordinary atom like `$x$'\thinspace;\cr Op&is a large operator atom like `$\sum$'\thinspace;\cr Bin&is a binary operation atom like `$+$'\thinspace;\cr Rel&is a relation atom like `$=$'\thinspace;\cr Open&is an opening atom like `$($'\thinspace;\cr Close&is a closing atom like `$)$'\thinspace;\cr Punct&is a punctuation atom like `$,$'\thinspace;\cr Inner&is an inner atom like `$1\over2$'\thinspace;\cr Over&is an overline atom like `$\overline x$'\thinspace;\cr Under&is an underline atom like `$\underline x$'\thinspace;\cr Acc&is an accented atom like `$\hat x$'\thinspace;\cr Rad&is a radical atom like `$\sqrt2$'\thinspace;\cr Vcent&is a vbox to be centered, produced by |\vcenter|.\cr }$$ \ddanger An atom's nucleus, superscript, and subscript are called its {\sl ^{fields}}, and there are four possibilities for each of these fields. A field can be\enddanger \smallskip \item\bull empty; \item\bull a math symbol (specified by family and position number); \item\bull a box; or \item\bull a math list. \smallskip\noindent For example, the Close atom $)^{\overline{n+1}}$ considered above has an empty subscript field; its nucleus is the symbol `$)$', which is character~\hex{28} of family~0 if the conventions of plain \TeX\ are in force; and its superscript field is the math list $\overline{n+1}$. The latter math list consists of an Over atom whose nucleus is the math list $n+1$; and that math list, in turn, consists of three atoms of types Ord, Bin, Ord. \ddanger You can see \TeX's view of a math list by typing ^|\showlists| in math mode. ^^{internal list format} For example, after `|$(x_i+y)^{\overline{n+1}}\showlists|' your log file gets the following curious data: \begindisplay |\mathopen|\cr |.\fam0 (|\cr |\mathord|\cr |.\fam1 x|\cr |_\fam1 i|\cr \noalign{\penalty-500} |\mathbin|\cr |.\fam0 +|\cr \noalign{\penalty-500} |\mathord|\cr |.\fam1 y|\cr \noalign{\penalty-500} |\mathclose|\cr |.\fam0 )|\cr |^\overline|\cr |^.\mathord|\cr |^..\fam1 n|\cr |^.\mathbin|\cr |^..\fam0 +|\cr |^.\mathord|\cr |^..\fam0 1|\cr \enddisplay In our previous experiences with |\showlists| we observed that there can be boxes within boxes, and that each line in the log file is prefixed by dots to indicate its position in the hierarchy. Math lists have a slightly more complex structure; therefore a dot is used to denote the nucleus of an atom, a~`|^|' is used for the superscript field, and a~`|_|' is used for the subscript field. Empty fields are not shown. Thus, for example, the Ord atom~$x_i$ is represented here by three lines `|\mathord|', `|.\fam1 x|', and `|_\fam1 i|'. \ddanger Certain kinds of atoms carry additional information besides their nucleus, subscript, and superscript fields: An Op atom will be marked `^|\limits|' or `^|\nolimits|' if the normal ^|\displaylimits| convention has been overridden; a Rad atom contains a delimiter field to specify what radical sign is to be used; and an Acc atom contains the family and character codes of the accent symbol. \ddanger When you say ^|\hbox||{...}| in math mode, an Ord atom is placed on the current math list, with the hbox as its nucleus. Similarly, ^|\vcenter||{...}| produces a Vcent atom whose nucleus is a box. But in most cases the nucleus of an atom will be either a symbol or a math list. You can experiment with |\showlists| to discover how other things like fractions and mathchoices are represented internally. \ddanger Chapter~26 contains complete details of how math lists are constructed. As soon as math mode ends (i.e., when the closing `|$|' occurs), \TeX\ dismantles the current math list and converts it into a horizontal list. The rules for this conversion are spelled out in Appendix~G\null. You can see ``before and after'' representations of such math typesetting by ending a formula with `|\showlists$\showlists|'; the first |\showlists| will display the math list, and the second will show the (possibly complex) horizontal list that is manufactured from it. \endchapter The learning time is short. A few minutes gives the general flavor, and typing a page or two of a paper generally uncovers most of the misconceptions. \author ^{KERNIGHAN} and ^{CHERRY}, {\sl A System for % Typesetting Mathematics\/} (1975) % in {\sl Communications of the ACM\/} p152 \bigskip Within a few hours (a few days at most) a typist with no math or typesetting experience can be taught to input even the most complex equations. \author PETER J. ^{BOEHM}, {\sl Software and Hardware Considerations % for a\break Technical Typesetting System\/} (1976) % in {\sl IEEE Transactions on Professional Communication\/} PC-19, pp15--19 \eject \beginchapter Chapter 18. Fine Points of\\Mathematics Typing We have discussed most of the facilities needed to construct math formulas, but there are several more things a good mathematical typist will want to watch for. After you have typed a dozen or so formulas using the basic ideas of Chapters 16 and~17, you will find that it's easy to visualize the final appearance of a mathematical expression as you type it. And once you have gotten to that level, there's only a little bit more to learn before you are producing formulas as beautiful as any the world has ever seen; tastefully applied touches of \TeX nique will add a professional polish that works wonders for the appearance and readability of the books and papers that you type. This chapter talks about such tricks, and it also fills in a few gaps by mentioning some aspects of math that didn't fit comfortably into~Chapters~16~and~17. \subsection Punctuation. When a formula is followed by a ^{period}, ^{comma}, ^{semicolon}, ^{colon}, ^{question mark}, ^{exclamation point}, etc., put the ^{punctuation} {\sl after\/} the |$|, when the formula is in the text; but put the punctuation {\sl before\/} the |$$| when the formula is displayed. For example, \begintt If $x<0$, we have shown that $$y=f(x).$$ \endtt \TeX's spacing rules within paragraphs work best when the ^{punctuation marks} are not considered to be part of the formulas. Similarly, don't ever type anything like \begintt for $x = a, b$, or $c$. \endtt It should be \begintt for $x = a$, $b$, or $c$. \endtt (Better yet, use a ^{tie}: `|or~$c$|'.) \ The reason is that \TeX\ will typeset expression `|$x|~|=|~|a,|~|b$|' as a single formula, so it will put a ``^{thin space}'' between the comma and the $b$. This space will not be the same as the space that \TeX\ puts after the comma {\sl after\/} the $b$, since spaces between words are always bigger than thin spaces. Such unequal spacing looks bad, but when you type things right the spacing will look good. Another reason for not typing `|$x| |=| |a,| |b$|' is that it inhibits the possibilities for breaking lines in a paragraph: \TeX\ will never break at the space between the comma and the |b| because breaks after commas in formulas are usually wrong. For example, in the equation `|$x|~|=|~|f(a,|~|b)$|' we certainly don't want to put `$x=f(a,$' on one line and `$b)$' on the next. Thus, when typing formulas in the text of a paragraph, keep the math properly segregated: Don't take operators like $-$ and $=$ outside of the |$|'s, and keep commas inside the formula if they are truly part of the formula. But if a comma or period or other punctuation mark belongs linguistically to the sentence rather than to the formula, leave it outside the |$|'s. \exercise Type this: $R(n,t)=O(t^{n/2})$, as $t\to0^+$. \answer |$R(n,t)=O(t^{n/2})$, as $t\to0^+$.| \ (N.B.: `|O(|', not `|0('|.) \danger Some mathematical styles insert a bit of extra space around formulas to separate them from the text. For example, when copy is being produced on an ordinary typewriter that doesn't have italic letters, the best technical typists have traditionally put an extra blank space before and after each formula, because this provides a useful visual distinction. You might find it helpful to think of each |$| as a symbol that has the potential of adding a little space to the printed output; then the rule about excluding sentence punctuation from formulas may be easier to remember. \ddanger \TeX\ does, in fact, insert additional ^{space} before and after each formula; the amount of such space is called ^|\mathsurround|, which is a \-valued parameter. For example, if you set |\mathsurround=1pt|, each formula will effectively be 2~points wider ($1\pt$ at each side): $$\baselineskip13pt\halign{\indent\mathsurround=#pt For $x=a$, $b$, or $c$.\hfil&\quad(#)\hfil\cr 1&|\mathsurround=1pt|\cr 0&|\mathsurround=0pt|\cr}$$ This extra space will disappear into the left or right margin if the formula occurs at the beginning or end of a line. The value of\/ |\mathsurround| that is in force when \TeX\ reads the closing |$| of a formula is used at both left and right of that formula. Plain \TeX\ takes |\mathsurround=0pt|, so you won't see any extra space unless you are using some other format, or unless you change |\mathsurround| yourself. \subsection Non-italic letters in formulas. The names of algebraic variables are usually italic or Greek letters, but common mathematical functions like `log' are always set in ^{roman type}. The best way to deal with such constructions is to make use of the following 32~control sequences (all of which are defined in plain \TeX\ format, see Appendix~B): \begintt \arccos \cos \csc \exp \ker \limsup \min \sinh \arcsin \cosh \deg \gcd \lg \ln \Pr \sup \arctan \cot \det \hom \lim \log \sec \tan \arg \coth \dim \inf \liminf \max \sin \tanh \endtt ^^|\arccos| ^^|\cos| ^^|\csc| ^^|\exp| ^^|\ker| ^^|\limsup| ^^|\min| ^^|\sinh| ^^|\arcsin| ^^|\cosh| ^^|\deg| ^^|\gcd| ^^|\lg| ^^|\ln| ^^|\Pr| ^^|\sup| ^^|\arctan| ^^|\cot| ^^|\det| ^^|\hom| ^^|\lim| ^^|\log| ^^|\sec| ^^|\tan| ^^|\arg| ^^|\coth| ^^|\dim| ^^|\inf| ^^|\liminf| ^^|\max| ^^|\sin| ^^|\tanh| These control sequences lead to roman type with appropriate spacing:\def\sep{&\hskip-1em} \beginlongmathdemo \it Input\sep\it Output\cr \noalign{\vskip2pt} |$\sin2\theta=2\sin\theta\cos\theta$|\sep\sin2\theta=2\sin\theta\cos\theta\cr |$O(n\log n\log\log n)$|\sep O(n\log n\log\log n)\cr |$\Pr(X>x)=\exp(-x/\mu)$|\sep\Pr(X>x)=\exp(-x/\mu)\cr |$$\max_{1\le n\le m}\log_2P_n$$|\sep \displaystyle{\max_{1\le n\le m}\log_2P_n}\cr \noalign{\vskip2pt} |$$\lim_{x\to0}{\sin x\over x}=1$$|\sep\displaystyle{\lim_{x\to0} {\sin x\over x}=1}\cr \endmathdemo ^^|\mu| The last two formulas, which are displays, show that some of the special control sequences are treated by \TeX\ as ``large operators'' with limits just like $\sum$: The subscript on |\max| is not treated like the subscript on |\log|. Subscripts and superscripts will become limits when they are attached to |\det|, |\gcd|, |\inf|, |\lim|, |\liminf|, |\limsup|, |\max|, |\min|, |\Pr|, and |\sup|, in display style. \exercise Express the following display in plain \TeX\ language, using `^|\nu|' for `$\nu$': $$p_1(n)=\lim_{m\to\infty}\sum_{\nu=0}^\infty \bigl(1-\cos^{2m}(\nu!^n\pi/n)\bigr).$$ \answer |$$p_1(n)=\lim_{m\to\infty}\sum_{\nu=0}^\infty|\parbreak | \bigl(1-\cos^{2m}(\nu!^n\pi/n)\bigr).$$|\par \smallskip\noindent $\bigl[$Mathematicians may enjoy interpreting this formula; cf.~G.~H. ^{Hardy}, {\sl Messenger of Mathematics\/ \bf35} (1906), 145--146.$\bigr]$ \danger If you need roman type for some mathematical function or operator that isn't included in plain \TeX's list of~32, it is easy to define a new control sequence by mimicking the definitions in Appendix~B\null. Or, if you need roman type just for a ``one shot'' use, it is even easier to get what you want by switching to ^|\rm| type, as follows: \beginlongmathdemo |$\sqrt{{\rm Var}(X)}$|&\sqrt{{\rm Var}(X)}\cr |$x_{\rm max}-x_{\rm min}$|&x_{\rm max}-x_{\rm min}\cr |${\rm LL}(k)\Rightarrow{\rm LR}(k)$|&{\rm LL}(k)\Rightarrow{\rm LR}(k)\cr |$\exp(x+{\rm constant})$|&\exp(x+{\rm constant})\cr |$x^3+{\rm lower\ order\ terms}$|&x^3+{\rm lower\ order\ terms}\cr \endmathdemo Notice the uses of `|\|\]' ^^{control space} in the last case; without them, the result would have been `$x^3+{\rm lower order terms}$', because ordinary blank ^{spaces} are ignored in math mode. \danger You can also use ^|\hbox| instead of\/ |\rm| to get roman letters into formulas. For example, four of the last five formulas can be generated by \beginlongmathdemo |$\sqrt{\hbox{Var}(X)}$|&\sqrt{\hbox{Var}(X)}\cr |$\hbox{LL}(k)\Rightarrow\hbox{LR}(k)$|&\hbox{LL}(k)\Rightarrow\hbox{LR}(k)\cr |$\exp(x+\hbox{constant})$|&\exp(x+\hbox{constant})\cr |$x^3+\hbox{lower order terms}$|&x^3+\hbox{lower order terms}\cr \endmathdemo In this case `|\|\]' isn't necessary, because the material in an |\hbox| is processed in horizontal mode, when spaces are significant. But such uses of\/ |\hbox| have two disadvantages: \ (1)~The contents of the box will be typeset in the same size, whether or not the box occurs as a subscript; for example, `|$x_{\hbox{max}}$|' yields `$x_{\hbox{max}}$'. \ (2)~The font that's used inside |\hbox| will be the ``^{current font},'' so it might not be roman. For example, if you are typesetting the statement of some theorem that is in slanted type, and if that theorem refers to `|$\sqrt{\hbox{Var}(X)}$|', you will get the unintended result `{\sl$\sqrt{\hbox{Var}(X)}$}'. In order to make sure that an |\hbox| uses roman type, you need to specify |\rm|, e.g., `|$\sqrt{\hbox{\rm Var}(X)}$|'; and then the |\hbox| serves no purpose. We will see later, however, that |\hbox| can be very useful in displayed formulas. \ddangerexercise When the displayed formula `|$$\lim_{n\to\infty}x_n {\rm\ exists} \iff|\break |\limsup_{n\to\infty}x_n = \liminf_{n\to\infty}x_n.$$|' is typeset with the standard macros of plain \TeX, you get $$\lim_{n\to\infty}x_n{\rm\ exists}\iff \limsup_{n\to\infty}x_n=\liminf_{n\to\infty}x_n.$$ But some people prefer a different notation: Explain how you could change the definitions of\/ ^|\limsup| and ^|\liminf| so that the display would be $$ \def\limsup{\mathop{\overline{\rm lim}}} \def\liminf{\mathop{\underline{\rm lim}}} \lim_{n\to\infty}x_n{\rm\ exists}\iff \limsup_{n\to\infty}x_n=\liminf_{n\to\infty}x_n.$$ \answer |\def\limsup{\mathop{\overline{\rm lim}}}|\parbreak |\def\liminf{\mathop{\underline{\rm lim}}}|\par \smallskip\noindent [Notice that the limits `$n\to\infty$' appear at different levels, in both of the displays, because `sup' and the underbar descend below the baseline. It is possible to unify the limit positions by using ^{phantoms}, as explained later in this chapter. For example, \begintt \def\limsup{\mathop{\vphantom{\underline{}}\overline{\rm lim}}} \endtt would give lower limits in the same position as |\liminf|.] \danger The word `mod' is also generally set in roman type, when it occurs in formulas; but this word needs more care, because it is used in two different ways that require two different treatments. Plain \TeX\ provides two different control sequences, ^|\bmod| and ^|\pmod|, for the two cases: |\bmod| is to be used when `mod' is a ^{binary operation} (i.e., when it occurs between two quantities, like a plus sign usually does), and |\pmod| is to be used when `mod' occurs parenthetically at the end of a formula. For example, \beginmathdemo |$\gcd(m,n)=\gcd(n,m\bmod n)$|&\gcd(m,n)=\gcd(n,m\bmod n)\cr |$x\equiv y+1\pmod{m^2}$|&x\equiv y+1\pmod{m^2}\cr \endmathdemo The `|b|' in `|\bmod|' stands for ``binary''; the `|p|' in `|\pmod|' stands for ``parenthesized.'' Notice that |\pmod| inserts its own parentheses; the quantity that appears after `mod' in the parentheses should be enclosed in braces, if it isn't a single symbol. \dangerexercise What did poor B. L. ^{User} get when he typed `|$x\equiv0 (\pmod y^n)$|'\thinspace? \answer $x\equiv0(\pmod y^n)$. He should have typed `|$x\equiv0\pmod{y^n}$|'. \dangerexercise Explain how to produce \lower12pt\null\ $\smash{\displaystyle{n\choose k}\equiv{\lfloor n/p\rfloor\choose \lfloor k/p\rfloor}{n\bmod p\choose k\bmod p}\pmod p.}$ \answer |$${n\choose k}\equiv{\lfloor n/p\rfloor\choose|\parbreak | \lfloor k/p\rfloor}{n\bmod p\choose k\bmod p}\pmod p.$$| \danger The same mechanism that works for roman type in formulas can be used to get other styles of type as well. For example, ^|\bf| yields ^{boldface}: \beginmathdemo |$\bf a+b=\Phi_m$|&\bf a+b=\Phi_m\cr \endmathdemo Notice that whole formula didn't become emboldened in this example; the `$+$' and `$=$' stayed the same. Plain \TeX\ sets things up so that commands like |\rm| and |\bf| will affect only the uppercase letters |A|~to~|Z|, the lowercase letters |a|~to~|z|, the digits |0|~to~|9|, the uppercase Greek letters |\Gamma| to~|\Omega|, and math ^{accents} like ^|\hat| and ^|\tilde|. Incidentally, no braces were used in this example, because |$|'s have the effect of grouping; |\bf| changes the current font, but the change is local, so it does not affect the font that was current outside the formula. \ddanger The bold fonts available in plain \TeX\ are ``bold roman,'' rather than ``bold italic,'' because the latter are rarely needed. However, \TeX\ could readily be set up to make use of bold math italics, if desired (see Exercise 17.\bmiexno). A more extensive set of math fonts would also include ^{script}, ^{Fraktur}, and ``^{blackboard bold}'' styles; plain \TeX\ doesn't have these, but other formats like \AmSTeX\ do. ^^{AMS-TeX} ^^{German black letters} \danger Besides |\rm| and |\bf|, you can say ^|\cal| in formulas to get uppercase letters in a ``^{calligraphic}'' style. For example, `|$\cal A$|' produces `$\cal A$' and `|$\cal Z$|' produces `$\cal Z$'. But beware: This works only with the letters |A| to |Z|; you'll get weird results if you apply |\cal| to lowercase or Greek letters. \danger There's also ^|\mit|, which stands for ``^{math italic}.'' This affects ^{uppercase Greek}, so that you get $\mit(\Gamma,\Delta,\Theta,\Lambda,\Xi,\Pi,\Sigma,\Upsilon,\Phi,\Psi,\Omega)$ instead of $(\Gamma,\ldots,\Omega)$. When~|\mit| is in effect, the ordinary letters |A| to |Z| and |a| to |z| are not changed; they are set in italics as usual, because they ordinarily come from the math italic font. Conversely, uppercase Greek letters and math accents are unaffected by |\rm|, because they ordinarily come from the roman font. Math accents should not be used when the |\mit| family has been selected, because the math italic font contains no accents. \dangerexercise Type the formula $\bf\bar x^{\rm T}Mx={\rm0}\iff x=0$, using as few keystrokes as possible. ^^{boldface numbers in math} \ (The first `0' is roman, the second is bold. The superscript `T' is roman.) \answer |$\bf\bar x^{\rm T}Mx={\rm0}\iff x=0$|. \ (If you typed a space between |\rm| and~|0|, you wasted a keystroke; but don't feel guilty about it.) \dangerexercise Figure out how to typeset `$S\subseteq\mit\Sigma\iff S\in\cal S$'. \answer |$S\subseteq{\mit\Sigma}\iff S\in{\cal S}$|. In this case the braces are redundant and could be eliminated; but you shouldn't try to do {\sl everything\/} with fewest keystrokes, or you'll outsmart yourself some day. \danger Plain \TeX\ also allows you to type ^|\it|, ^|\sl|, or ^|\tt|, if you want text italic, slanted, or typewriter letters to occur in a math formula. However, these fonts are available only in text size, so you should not try to use them in subscripts. \danger If you're paying attention, you probably wonder why both |\mit| and |\it| are provided; the answer is that |\mit| is ``math italic'' (which is normally best for formulas), and |\it| is ``^{text italic}'' (which is normally best for running text). \beginmathdemo |$This\ is\ math\ italic.$|&This\ is\ math\ italic.\cr |{\it This is text italic.}|&\hbox{\it This is text italic.}\cr \endmathdemo The math italic letters are a little wider, and the spacing is different; this works better in most formulas, but it fails spectacularly when you try to type certain italic words like `$different$' using math mode (`|$different$|'). A wide `$f$' is usually desirable in formulas, but it is undesirable in text. Therefore wise typists use |\it| in a math formula that is supposed to contain an actual italic word. Such cases almost never occur in classical mathematics, but they are common when ^{computer programs} are being typeset, since programmers often use multi-letter ``^{identifiers}'': \beginmathdemo |$\it last:=first$|&\it last:=first\cr |$\it x\_coord(point\_2)$|&\it x\_coord(point\_2)\cr \endmathdemo The first of these examples shows that \TeX\ recognizes the ^{ligature} `{\it fi\/}' when text italic occurs in a math formula; the other example illustrates the use of short ^{underlines} to break up identifier names. ^^{control-underline} When the author typeset this manual, he used `|$\it SS$|' to refer to style~$\SS$, since `|$SS$|' makes the $S$'s too far apart: $SS$. \dangerexercise What plain \TeX\ commands will produce the following display? $$\tenmath {\it available}+\sum_{i=1}^n\max\bigl({\it full}(i),{\it reserved}(i)\bigr) ={\it capacity}.$$ \answer |$${\it available}+\sum_{i=1}^n\max\bigl({\it full}(i),|\parbreak |{\it reserved}(i)\bigr)={\it capacity}.$$| \smallskip\noindent [If\/ |\it| had been used throughout the formula, the subscript~$i$ and superscript~$n$ would have caused error messages saying `^|\scriptfont| |4| |is| |undefined|', since plain \TeX\ makes |\it| available only in text size.] \ddangerexercise How would you go about typesetting the following computer program, using the macros of plain \TeX? $$\vbox{\let\par=\endgraf \obeylines\sfcode`;=3000 {\bf for $j:=2$ step $1$ until $n$ do} \quad {\bf begin} ${\it accum}:=A[j]$; $k:=j-1$; $A[0]:=\it accum$; \quad {\bf while $A[k]>\it accum$ do} \qquad {\bf begin} $A[k+1]:=A[k]$; $k:=k-1$; \qquad {\bf end}; \quad $A[k+1]:=\it accum$; \quad {\bf end}. }$$ \answer |{\obeylines \sfcode`;=3000|^^|\sfcode|\parbreak |{\bf for $j:=2$ step $1$ until $n$ do}|\parbreak |\quad {\bf begin} ${\it accum}:=A[j]$; $k:=j-1$; $A[0]:=\it accum$;|\parbreak |\quad {\bf while $A[k]>\it accum$ do}|\parbreak |\qquad {\bf begin} $A[k+1]:=A[k]$; $k:=k-1$;|\parbreak |\qquad {\bf end};|\parbreak |\quad $A[k+1]:=\it accum$;|\parbreak |\quad {\bf end}.\par}|\par \smallskip\noindent [This is something like the ``poetry'' example in Chapter~14, but much more difficult. Some manuals of style say that ^{punctuation} should inherit the font of the preceding character, so that three kinds of semicolons should be typeset; e.g., these experts recommend `$k:=j-1$; \ $A[0]:={}${\it accum;} \ {\bf end;}'. The author heartily disagrees.] \subsection Spacing between formulas. ^{Displays} often contain more than one formula; for example, an equation is frequently accompanied by a ^{side condition}: $$F_n=F_{n-1}+F_{n-2},\qquad n\ge2.$$ In such cases you need to tell \TeX\ how much space to put after the comma, because \TeX's normal spacing conventions would bunch things together; without special precautions you would get $$F_n=F_{n-1}+F_{n-2}, n\ge2.$$ The traditional hot-metal technology for printing has led to some ingrained standards for situations like this, based on what printers call a ``^{quad}'' of space. Since these standards seem to work well in practice, \TeX\ makes it easy for you to continue the tradition: When you type `^|\quad|' in plain \TeX\ format, you get a printer's quad of space in the horizontal direction. Similarly, `^|\qquad|' gives you a double quad (twice as much); this is the normal spacing for situations like the $F_n$ example above. Thus, the recommended procedure is to type \begintt $$ F_n = F_{n-1} + F_{n-2}, \qquad n \ge 2. $$ \endtt It is perhaps worth reiterating that \TeX\ ignores all the spaces in math mode (except, of course, the space after `|\qquad|', which is needed to distinguish between `|\qquad|~|n|' and `|\qquadn|'); so the same result would be obtained if you were to leave out all but one space: \begintt $$F_n=F_{n-1}+F_{n-2},\qquad n\ge2.$$ \endtt Whenever you want spacing that differs from the normal conventions, you must specify it explicitly by using control sequences such as |\quad| and |\qquad|. \danger A quad used to be a square piece of blank type, $1\em$ wide and $1\em$ tall---approximately the size of a capital M, as explained in Chapter~10. This tradition has not been fully retained: The control sequence |\quad| in plain \TeX\ is simply an abbreviation for `|\hskip|~|1|^|em||\relax|', so \TeX's quad has width but no height. \danger You can use |\quad| in text as well as in formulas; for example, Chapter~14 illustrates how |\quad| applies to poetry. When |\quad| appears in a formula it stands for one~em in the current text font, independent of the current math size or style or family. Thus, for example, |\quad| is just as wide in a subscript as it is on the main line of a formula. Sometimes a careless author will put two formulas next to each other in the text of a paragraph. For example, you might find a sentence like this: \begindisplay The ^{Fibonacci} numbers satisfy $F_n=F_{n-1}+F_{n-2}$, \ $n\ge2$. \enddisplay Everybody who teaches proper ^{mathematical} ^{style} is agreed that formulas ought to be separated by words, not just by commas; the author of that sentence should at least have said `for $n\ge2$', not simply `$n\ge2$'. But alas, such lapses are commonplace, and many prominent mathematicians are hopelessly addicted to clusters of formulas. If we are not allowed to change their writing style, we can at least insert extra space where they neglected to insert an appropriate word. An additional interword space generally works well in such cases; for example, the sentence above was typeset thus: \begintt ... $F_n=F_{n-1}+F_{n-2}$, \ $n\ge2$.}$$ \endtt The `|\|\]' ^^{control space} here gives a visual separation that partly compensates for the bad style. \exercise Put the following paragraph into \TeX\ form, treating punctuation and spacing carefully; also insert ^{ties} to prevent bad line breaks. \begindisplay\baselineskip13pt \vbox{\raggedright\hsize=310pt\parindent=0pt Let $H$~be a Hilbert space, \ $C$~a closed bounded convex subset of~$H$, \ $T$~a nonexpansive self map of~$C$. Suppose that as $n\to\infty$, \ $a_{n,k}\to0$ for each~$k$, and $\gamma_n=\sum_{k=0}^\infty(a_{n,k+1}-a_{n,k})^+\to0$. Then for each $x$~in~$C$, \ $A_nx=\sum_{k=0}^\infty a_{n,k}T^kx$ converges weakly to a fixed point of~$T$. } % taken from Bull. AMS 82 (1976), p 959; chosen by AMS in '78 for demo \enddisplay \answer |Let $H$~be a Hilbert space, \ $C$~a closed bounded convex subset of~$H$, \ $T$~a nonexpansive self map of~$C$. Suppose that as $n\to\infty$, \ $a_{n,k}\to0$ for each~$k$, and $\gamma_n=\sum_{k=0}^\infty(a_{n,k+1}-|\allowbreak|a_{n,k})^+\to0$. Then for each $x$~in~$C$, \ $A_nx=\sum_{k=0}^\infty a_{n,k}T^kx$ converges weakly to a fixed point of~$T$.|\par [If any mathematicians are reading this, they might either appreciate or resent the following attempt to edit the given paragraph into a more acceptable style: ``% Let $C$~be a closed, bounded, convex subset of a Hilbert space~$H$, and let $T$~be a nonexpansive self map of~$C$. Suppose that as $n\to\infty$, we have $a_{n,k}\to0$ for each~$k$, and $\gamma_n=\sum_{k=0}^\infty(a_{n,k+1}-a_{n,k})^+\to0$. Then for each $x$~in~$C$, the infinite sum $A_nx=\sum_{k=0}^\infty a_{n,k}T^kx$ converges weakly to a fixed point of~$T$.''] \subsection Spacing within formulas. Chapter 16 says that \TeX\ does automatic ^{spacing} of math formulas so that they look right, and this is almost true. But occasionally you must give \TeX\ some help. The number of possible math formulas is vast, and \TeX's spacing rules are rather simple, so it is natural that exceptions should arise. Of course, it is desirable to have fine units of spacing for this purpose, instead of the big chunks that arise from |\|\], |\quad| and |\qquad|. The basic elements of space that \TeX\ puts into formulas are called {\sl ^{thin spaces}}, {\sl ^{medium spaces}}, and {\sl ^{thick spaces}}. In order to get a feeling for these units, let's take a look at the $F_n$ example again: Thick spaces occur just before and after the = sign, and also before and after the $\ge$\thinspace; medium spaces occur just before and after the $+$ sign. Thin spaces are slightly smaller, but noticeable; it's a thin space that makes the difference between `loglog' and `$\log\log$'. The normal space between words of a paragraph is approximately equal to two thin spaces. \TeX\ inserts thin spaces, medium spaces, and thick spaces into formulas automatically, but you can add your own spacing whenever you want to, by using the control sequences ^^|\,|^^|\!|^^|\;|^^|\>| $$\halign{\indent#\hfil&\quad#\hfil\cr |\,|&thin space \ (normally 1/6 of a quad);\cr |\>|&medium space \ (normally 2/9 of a quad);\cr |\;|&thick space \ (normally 5/18 of a quad);\cr |\!|&negative thin space \ (normally $-1/6$ of a quad).\cr}$$ In most cases you can rely on \TeX's spacing while you are typing a manuscript, and you'll want to insert or delete space with these four control sequences only in rare circumstances after you see what comes out. \ddanger We observed a minute ago that |\quad| spacing does not change with the style of formula, nor does it depend on the math font families that are being used. But thin spaces, medium spaces, and thick spaces do get bigger and smaller as the size of type gets bigger and smaller; this is because they are defined in terms of ^\, a~special brand of glue intended for math spacing. You specify \ just as if it were ordinary glue, except that the units are given in terms of `^|mu|' (math units) instead of~|pt| or~|cm| or something else. For example, Appendix~B contains the definitions \begintt \thinmuskip = 3mu \medmuskip = 4mu plus 2mu minus 4mu \thickmuskip = 5mu plus 5mu \endtt ^^|\thinmuskip|^^|\medmuskip|^^|\thickmuskip| and this defines the thin, medium, and thick spaces that \TeX\ inserts into formulas. According to these specifications, thin spaces in plain \TeX\ do not stretch or shrink; medium spaces can stretch a little, and they can shrink to zero; thick spaces can stretch a lot, but they never shrink. \ddanger There are 18 mu to an em, where the em is taken from family~2 (the math symbols family). In other words, ^|\textfont|~|2| defines the em value for |mu| in display and text styles; ^|\scriptfont|~|2| defines the em for script size material; and ^|\scriptscriptfont|~|2| defines it for scriptscript size. \ddanger You can insert math glue into any formula just by giving the command `^|\mskip|\'. For example, `|\mskip 9mu plus 2mu|' inserts one half em of space, in the current size, together with some stretchability. Appendix~B defines `|\,|' to be an abbreviation for `|\mskip\thinmuskip|'. Similarly, you can use the command `^|\mkern|' when there is no stretching or shrinking; `|\mkern18mu|' gives one em of horizontal space in the current size. \TeX\ insists that |\mskip| and |\mkern| be used only with |mu|; conversely, ^|\hskip| and ^|\kern| (which are also allowed in formulas) must never give units in |mu|. Formulas involving ^{calculus} look best when an extra thin space appears before $dx$ ^^{dx} or~$dy$ or~$d\,$whatever; but \TeX\ doesn't do this automatically. Therefore a well-trained typist will remember to insert `|\,|' in examples like the following: \beginmathdemo \it Input&\it Output\cr \noalign{\vskip2pt} |$\int_0^\infty f(x)\,dx$|&\int_0^\infty f(x)\,dx\cr |$y\,dx-x\,dy$|&y\,dx-x\,dy\cr |$dx\,dy=r\,dr\,d\theta$|&dx\,dy=r\,dr\,d\theta\cr |$x\,dy/dx$|&x\,dy/dx\cr \endmathdemo Notice that no `|\,|' was desirable after the `|/|' in the last example. Similarly, there's no need for `|\,|' in cases like \begindisplaymathdemo |$$\int_1^x{dt\over t}$$|&\int_1^x{dt\over t}\cr \endmathdemo since the $dt$ appears all by itself in the numerator of a fraction; this detaches it visually from the rest of the formula. \exercise Explain how to handle the display $$\int_0^\infty{t-ib\over t^2+b^2}e^{iat}\,dt=e^{ab}E_1(ab),\qquad a,b>0.$$ \answer |$$\int_0^\infty{t-ib\over t^2+b^2}e^{iat}\,dt=|\parbreak | e^{ab}E_1(ab),\qquad a,b>0.$$| \danger When physical ^{units} appear in a formula, they should be set in roman type and separated from the preceding material by a thin space: \beginmathdemo |$55\rm\,mi/hr$|&55\rm\,mi/hr\cr |$g=9.8\rm\,m/sec^2$|&g=9.8\rm\,m/sec^2\cr |$\rm1\,ml=1.000028\,cc$|&\rm1\,ml=1.000